From b603a10002b98a2f7d9132a0c8dc21fec3cc7339 Mon Sep 17 00:00:00 2001
From: Ray Wang <raywang777@foxmail.com>
Date: Fri, 19 Jan 2024 13:58:56 +0800
Subject: [PATCH 01/12] implement fast guided filter

---
 modules/ximgproc/doc/ximgproc.bib             |  7 ++
 .../include/opencv2/ximgproc/edge_filter.hpp  | 20 +++--
 modules/ximgproc/perf/perf_guided_filter.cpp  |  7 +-
 modules/ximgproc/src/guided_filter.cpp        | 88 +++++++++++++++----
 modules/ximgproc/test/test_guided_filter.cpp  | 50 +++++++++++
 5 files changed, 144 insertions(+), 28 deletions(-)

diff --git a/modules/ximgproc/doc/ximgproc.bib b/modules/ximgproc/doc/ximgproc.bib
index f081f54d3ce..279bac1d115 100644
--- a/modules/ximgproc/doc/ximgproc.bib
+++ b/modules/ximgproc/doc/ximgproc.bib
@@ -54,6 +54,13 @@ @incollection{Kaiming10
   publisher={Springer}
 }
 
+@article{Kaiming15,
+  title={Fast guided filter},
+  author={He, Kaiming and Sun, Jian},
+  journal={arXiv preprint arXiv:1505.00996},
+  year={2015}
+}
+
 @inproceedings{Lee14,
   title={Outdoor place recognition in urban environments using straight lines},
   author={Lee, Jin Han and Lee, Sehyung and Zhang, Guoxuan and Lim, Jongwoo and Chung, Wan Kyun and Suh, Il Hong},
diff --git a/modules/ximgproc/include/opencv2/ximgproc/edge_filter.hpp b/modules/ximgproc/include/opencv2/ximgproc/edge_filter.hpp
index 82be7c71b7f..19b05451147 100644
--- a/modules/ximgproc/include/opencv2/ximgproc/edge_filter.hpp
+++ b/modules/ximgproc/include/opencv2/ximgproc/edge_filter.hpp
@@ -123,15 +123,15 @@ void dtFilter(InputArray guide, InputArray src, OutputArray dst, double sigmaSpa
 //////////////////////////////////////////////////////////////////////////
 //////////////////////////////////////////////////////////////////////////
 
-/** @brief Interface for realizations of Guided Filter.
+/** @brief Interface for realizations of (Fast) Guided Filter.
 
-For more details about this filter see @cite Kaiming10 .
+For more details about this filter see @cite Kaiming10 @cite Kaiming15 .
  */
 class CV_EXPORTS_W GuidedFilter : public Algorithm
 {
 public:
 
-    /** @brief Apply Guided Filter to the filtering image.
+    /** @brief Apply (Fast) Guided Filter to the filtering image.
 
     @param src filtering image with any numbers of channels.
 
@@ -153,11 +153,14 @@ channels then only first 3 channels will be used.
 @param eps regularization term of Guided Filter. \f${eps}^2\f$ is similar to the sigma in the color
 space into bilateralFilter.
 
-For more details about Guided Filter parameters, see the original article @cite Kaiming10 .
+@param scale subsample factor of Fast Guided Filter, use a scale less than 1 to speeds up computation
+with almost no visible degradation. (e.g. scale==0.5 shrinks the image by 2x inside the filter)
+
+For more details about (Fast) Guided Filter parameters, see the original articles @cite Kaiming10 @cite Kaiming15 .
  */
-CV_EXPORTS_W Ptr<GuidedFilter> createGuidedFilter(InputArray guide, int radius, double eps);
+CV_EXPORTS_W Ptr<GuidedFilter> createGuidedFilter(InputArray guide, int radius, double eps, double scale = 1.0);
 
-/** @brief Simple one-line Guided Filter call.
+/** @brief Simple one-line (Fast) Guided Filter call.
 
 If you have multiple images to filter with the same guided image then use GuidedFilter interface to
 avoid extra computations on initialization stage.
@@ -176,8 +179,11 @@ space into bilateralFilter.
 
 @param dDepth optional depth of the output image.
 
+@param scale subsample factor of Fast Guided Filter, use a scale less than 1 to speeds up computation
+with almost no visible degradation. (e.g. scale==0.5 shrinks the image by 2x inside the filter)
+
 @sa bilateralFilter, dtFilter, amFilter */
-CV_EXPORTS_W void guidedFilter(InputArray guide, InputArray src, OutputArray dst, int radius, double eps, int dDepth = -1);
+CV_EXPORTS_W void guidedFilter(InputArray guide, InputArray src, OutputArray dst, int radius, double eps, int dDepth = -1, double scale = 1.0);
 
 //////////////////////////////////////////////////////////////////////////
 //////////////////////////////////////////////////////////////////////////
diff --git a/modules/ximgproc/perf/perf_guided_filter.cpp b/modules/ximgproc/perf/perf_guided_filter.cpp
index 9a0058a5505..7649acea9b3 100644
--- a/modules/ximgproc/perf/perf_guided_filter.cpp
+++ b/modules/ximgproc/perf/perf_guided_filter.cpp
@@ -7,11 +7,11 @@ namespace opencv_test { namespace {
 
 CV_ENUM(GuideTypes, CV_8UC1, CV_8UC3, CV_32FC1, CV_32FC3);
 CV_ENUM(SrcTypes, CV_8UC1, CV_8UC3, CV_32FC1, CV_32FC3);
-typedef tuple<GuideTypes, SrcTypes, Size> GFParams;
+typedef tuple<GuideTypes, SrcTypes, Size, double> GFParams;
 
 typedef TestBaseWithParam<GFParams> GuidedFilterPerfTest;
 
-PERF_TEST_P( GuidedFilterPerfTest, perf, Combine(GuideTypes::all(), SrcTypes::all(), Values(sz1080p, sz2K)) )
+PERF_TEST_P( GuidedFilterPerfTest, perf, Combine(GuideTypes::all(), SrcTypes::all(), Values(sz1080p, sz2K), Values(1./1, 1./2, 1./3, 1./4)) )
 {
     RNG rng(0);
 
@@ -19,6 +19,7 @@ PERF_TEST_P( GuidedFilterPerfTest, perf, Combine(GuideTypes::all(), SrcTypes::al
     int guideType   = get<0>(params);
     int srcType     = get<1>(params);
     Size sz         = get<2>(params);
+    double scale    = get<3>(params);
 
     Mat guide(sz, guideType);
     Mat src(sz, srcType);
@@ -30,7 +31,7 @@ PERF_TEST_P( GuidedFilterPerfTest, perf, Combine(GuideTypes::all(), SrcTypes::al
     {
         int radius = rng.uniform(5, 30);
         double eps = rng.uniform(0.1, 1e5);
-        guidedFilter(guide, src, dst, radius, eps);
+        guidedFilter(guide, src, dst, radius, eps, -1, scale);
     }
 
     SANITY_CHECK_NOTHING();
diff --git a/modules/ximgproc/src/guided_filter.cpp b/modules/ximgproc/src/guided_filter.cpp
index 11fa0f6fcb0..9d8333aecba 100644
--- a/modules/ximgproc/src/guided_filter.cpp
+++ b/modules/ximgproc/src/guided_filter.cpp
@@ -128,7 +128,7 @@ class GuidedFilterImpl : public GuidedFilter
 {
 public:
 
-    static Ptr<GuidedFilterImpl> create(InputArray guide, int radius, double eps);
+    static Ptr<GuidedFilterImpl> create(InputArray guide, int radius, double eps, double scale);
 
     void filter(InputArray src, OutputArray dst, int dDepth = -1) CV_OVERRIDE;
 
@@ -136,10 +136,13 @@ class GuidedFilterImpl : public GuidedFilter
 
     int radius;
     double eps;
+    double scale;
     int h, w;
+    int hOriginal, wOriginal;
 
     vector<Mat> guideCn;
     vector<Mat> guideCnMean;
+    vector<Mat> guideCnOriginal;
 
     SymArray2D<Mat> covarsInv;
 
@@ -149,7 +152,7 @@ class GuidedFilterImpl : public GuidedFilter
 
     GuidedFilterImpl() {}
 
-    void init(InputArray guide, int radius, double eps);
+    void init(InputArray guide, int radius, double eps, double scale);
 
     void computeCovGuide(SymArray2D<Mat>& covars);
 
@@ -167,6 +170,16 @@ class GuidedFilterImpl : public GuidedFilter
         src.convertTo(dst, CV_32F);
     }
 
+    inline void subsample(Mat& src, Mat& dst)
+    {
+        resize(src, dst, Size(w, h), 0, 0, INTER_LINEAR);
+    }
+
+    inline void upsample(Mat& src, Mat& dst)
+    {
+        resize(src, dst, Size(wOriginal, hOriginal), 0, 0, INTER_LINEAR);
+    }
+
 private: /*Routines to parallelize boxFilter and convertTo*/
 
     typedef void (GuidedFilterImpl::*TransformFunc)(Mat& src, Mat& dst);
@@ -203,6 +216,20 @@ class GuidedFilterImpl : public GuidedFilter
         parallel_for_(pb.getRange(), pb);
     }
 
+    template<typename V>
+    void parSubsample(V &src, V &dst)
+    {
+        GFTransform_ParBody pb(*this, src, dst, &GuidedFilterImpl::subsample);
+        parallel_for_(pb.getRange(), pb);
+    }
+
+    template<typename V>
+    void parUpsample(V &src, V &dst)
+    {
+        GFTransform_ParBody pb(*this, src, dst, &GuidedFilterImpl::upsample);
+        parallel_for_(pb.getRange(), pb);
+    }
+
 private: /*Parallel body classes*/
 
     inline void runParBody(const ParallelLoopBody& pb)
@@ -582,7 +609,7 @@ void GuidedFilterImpl::ApplyTransform_ParBody::operator()(const Range& range) co
     {
         float *_g[4];
         for (int gi = 0; gi < gf.gCnNum; gi++)
-            _g[gi] = gf.guideCn[gi].ptr<float>(i);
+            _g[gi] = gf.guideCnOriginal[gi].ptr<float>(i);
 
         float *betaDst, *g, *a;
         for (int si = 0; si < srcCnNum; si++)
@@ -593,7 +620,7 @@ void GuidedFilterImpl::ApplyTransform_ParBody::operator()(const Range& range) co
                 a = alpha[si][gi].ptr<float>(i);
                 g = _g[gi];
 
-                add_mul(betaDst, a, g, gf.w);
+                add_mul(betaDst, a, g, gf.wOriginal);
             }
         }
     }
@@ -666,28 +693,42 @@ void GuidedFilterImpl::getWalkPattern(int eid, int &cn1, int &cn2)
     cn2 = wdata[6 * 2 * (gCnNum-1) + 6 + eid];
 }
 
-Ptr<GuidedFilterImpl> GuidedFilterImpl::create(InputArray guide, int radius, double eps)
+Ptr<GuidedFilterImpl> GuidedFilterImpl::create(InputArray guide, int radius, double eps, double scale)
 {
     GuidedFilterImpl *gf = new GuidedFilterImpl();
-    gf->init(guide, radius, eps);
+    gf->init(guide, radius, eps, scale);
     return Ptr<GuidedFilterImpl>(gf);
 }
 
-void GuidedFilterImpl::init(InputArray guide, int radius_, double eps_)
+void GuidedFilterImpl::init(InputArray guide, int radius_, double eps_, double scale_)
 {
     CV_Assert( !guide.empty() && radius_ >= 0 && eps_ >= 0 );
     CV_Assert( (guide.depth() == CV_32F || guide.depth() == CV_8U || guide.depth() == CV_16U) && (guide.channels() <= 3) );
+    CV_Assert( scale_ <= 1.0 );
 
     radius = radius_;
     eps = eps_;
+    scale = scale_;
 
-    splitFirstNChannels(guide, guideCn, 3);
-    gCnNum = (int)guideCn.size();
-    h = guideCn[0].rows;
-    w = guideCn[0].cols;
+    splitFirstNChannels(guide, guideCnOriginal, 3);
+    gCnNum = (int)guideCnOriginal.size();
+    hOriginal = guideCnOriginal[0].rows;
+    wOriginal = guideCnOriginal[0].cols;
+    h = int(hOriginal * scale);
+    w = int(wOriginal * scale);
+
+    parConvertToWorkType(guideCnOriginal, guideCnOriginal);
+    if (scale < 1.0)
+    {
+        guideCn.resize(gCnNum);
+        parSubsample(guideCnOriginal, guideCn);
+    }
+    else
+    {
+        guideCn = guideCnOriginal;
+    }
 
     guideCnMean.resize(gCnNum);
-    parConvertToWorkType(guideCn, guideCn);
     parMeanFilter(guideCn, guideCnMean);
 
     SymArray2D<Mat> covars;
@@ -712,7 +753,7 @@ void GuidedFilterImpl::computeCovGuide(SymArray2D<Mat>& covars)
 void GuidedFilterImpl::filter(InputArray src, OutputArray dst, int dDepth /*= -1*/)
 {
     CV_Assert( !src.empty() && (src.depth() == CV_32F || src.depth() == CV_8U) );
-    if (src.rows() != h || src.cols() != w)
+    if (src.rows() != hOriginal || src.cols() != wOriginal)
     {
         CV_Error(Error::StsBadSize, "Size of filtering image must be equal to size of guide image");
         return;
@@ -725,6 +766,11 @@ void GuidedFilterImpl::filter(InputArray src, OutputArray dst, int dDepth /*= -1
     vector<Mat>& srcCnMean = srcCn;
     split(src, srcCn);
 
+    if (scale < 1.0)
+    {
+        parSubsample(srcCn, srcCn);
+    }
+
     if (src.depth() != CV_32F)
     {
         parConvertToWorkType(srcCn, srcCn);
@@ -749,7 +795,13 @@ void GuidedFilterImpl::filter(InputArray src, OutputArray dst, int dDepth /*= -1
     parMeanFilter(beta, beta);
     parMeanFilter(alpha, alpha);
 
-    runParBody(ApplyTransform_ParBody(*this, alpha, beta));
+    if (scale < 1.0)
+    {
+        parUpsample(beta, beta);
+        parUpsample(alpha, alpha);
+    }
+
+    parallel_for_(Range(0, hOriginal), ApplyTransform_ParBody(*this, alpha, beta));
     if (dDepth != CV_32F)
     {
         for (int i = 0; i < srcCnNum; i++)
@@ -782,15 +834,15 @@ void GuidedFilterImpl::computeCovGuideAndSrc(vector<Mat>& srcCn, vector<Mat>& sr
 //////////////////////////////////////////////////////////////////////////
 
 CV_EXPORTS_W
-Ptr<GuidedFilter> createGuidedFilter(InputArray guide, int radius, double eps)
+Ptr<GuidedFilter> createGuidedFilter(InputArray guide, int radius, double eps, double scale)
 {
-    return Ptr<GuidedFilter>(GuidedFilterImpl::create(guide, radius, eps));
+    return Ptr<GuidedFilter>(GuidedFilterImpl::create(guide, radius, eps, scale));
 }
 
 CV_EXPORTS_W
-void guidedFilter(InputArray guide, InputArray src, OutputArray dst, int radius, double eps, int dDepth)
+void guidedFilter(InputArray guide, InputArray src, OutputArray dst, int radius, double eps, int dDepth, double scale)
 {
-    Ptr<GuidedFilter> gf = createGuidedFilter(guide, radius, eps);
+    Ptr<GuidedFilter> gf = createGuidedFilter(guide, radius, eps, scale);
     gf->filter(src, dst, dDepth);
 }
 
diff --git a/modules/ximgproc/test/test_guided_filter.cpp b/modules/ximgproc/test/test_guided_filter.cpp
index b8293da65d0..c2263ecdd3b 100644
--- a/modules/ximgproc/test/test_guided_filter.cpp
+++ b/modules/ximgproc/test/test_guided_filter.cpp
@@ -64,6 +64,16 @@ static Mat convertTypeAndSize(Mat src, int dstType, Size dstSize)
     return dst;
 }
 
+static double laplacianVariance(Mat src)
+{
+    Mat laplacian;
+    Laplacian(src, laplacian, CV_64F);
+    Scalar mean, stddev;
+    meanStdDev(laplacian, mean, stddev);
+    double variance = stddev.val[0] * stddev.val[0];
+    return variance;
+}
+
 class GuidedFilterRefImpl : public GuidedFilter
 {
     int height, width, rad, chNum;
@@ -350,6 +360,46 @@ TEST_P(GuidedFilterTest, accuracy)
     }
 }
 
+TEST_P(GuidedFilterTest, accuracyFastGuidedFilter)
+{
+    int radius = 8;
+    double eps = 1;
+
+    GFParams params = GetParam();
+    string guideFileName = get<1>(params);
+    string srcFileName = get<2>(params);
+    int guideCnNum = 3;
+    int srcCnNum = get<0>(params);
+
+    Mat guide = imread(getOpenCVExtraDir() + guideFileName);
+    Mat src = imread(getOpenCVExtraDir() + srcFileName);
+    ASSERT_TRUE(!guide.empty() && !src.empty());
+
+    Size dstSize(guide.cols, guide.rows);
+    guide = convertTypeAndSize(guide, CV_MAKE_TYPE(guide.depth(), guideCnNum), dstSize);
+    src = convertTypeAndSize(src, CV_MAKE_TYPE(src.depth(), srcCnNum), dstSize);
+    Mat outputRef;
+    ximgproc::guidedFilter(guide, src, outputRef, radius, eps);
+
+    for (double scale : {1./2, 1./3, 1./4}) {
+        Mat outputFastGuidedFilter;
+        ximgproc::guidedFilter(guide, src, outputFastGuidedFilter, radius, eps, -1, scale);
+
+        Mat guideNaiveDownsampled, srcNaiveDownsampled, outputNaiveDownsampled;
+        resize(guide, guideNaiveDownsampled, {}, scale, scale, INTER_LINEAR);
+        resize(src, srcNaiveDownsampled, {}, scale, scale, INTER_LINEAR);
+        ximgproc::guidedFilter(guideNaiveDownsampled, srcNaiveDownsampled, outputNaiveDownsampled, radius, eps);
+        resize(outputNaiveDownsampled, outputNaiveDownsampled, dstSize, 0, 0, INTER_LINEAR);
+
+        double laplacianVarianceFastGuidedFilter = laplacianVariance(outputFastGuidedFilter);
+        double laplacianVarianceNaiveDownsampled = laplacianVariance(outputNaiveDownsampled);
+        EXPECT_GT(laplacianVarianceFastGuidedFilter, laplacianVarianceNaiveDownsampled);
+
+        double normL2 = cv::norm(outputFastGuidedFilter, outputRef, NORM_L2) / guide.total();
+        EXPECT_LE(normL2, 1.0/48.0/scale);
+    }
+}
+
 TEST_P(GuidedFilterTest, smallParamsIssue)
 {
     GFParams params = GetParam();

From fb69ae38cc774916e6132409c8245698ebe0d310 Mon Sep 17 00:00:00 2001
From: catree <catree.catreus@outlook.com>
Date: Sun, 21 Apr 2024 13:57:39 +0200
Subject: [PATCH 02/12] Use "compute" instead of "get" term. Add some classical
 visual servo references.

---
 modules/tracking/doc/tracking.bib             | 40 +++++++++++++++++++
 .../include/opencv2/tracking/twist.hpp        | 11 ++---
 modules/tracking/src/twist.cpp                |  4 +-
 modules/tracking/test/test_twist.cpp          |  4 +-
 4 files changed, 50 insertions(+), 9 deletions(-)

diff --git a/modules/tracking/doc/tracking.bib b/modules/tracking/doc/tracking.bib
index 78ce6c32fa1..ce49bb4026e 100644
--- a/modules/tracking/doc/tracking.bib
+++ b/modules/tracking/doc/tracking.bib
@@ -76,3 +76,43 @@ @Article{Lukezic_IJCV2018
 journal={International Journal of Computer Vision},
 year={2018},
 }
+
+@article{chaumette:inria-00350283,
+  title={{Visual servo control, Part I: Basic approaches}},
+  author={Chaumette, Fran{\c c}ois and Hutchinson, S.},
+  url={https://inria.hal.science/inria-00350283},
+  journal={{IEEE Robotics and Automation Magazine}},
+  publisher={{Institute of Electrical and Electronics Engineers}},
+  volume={13},
+  number={4},
+  pages={82-90},
+  year={2006},
+  pdf={https://inria.hal.science/inria-00350283/file/2006_ieee_ram_chaumette.pdf},
+  hal_id={inria-00350283},
+  hal_version={v1},
+}
+
+@article{chaumette:inria-00350638,
+  title={{Visual servo control, Part II: Advanced approaches}},
+  author={Chaumette, Fran{\c c}ois and Hutchinson, S.},
+  url={https://inria.hal.science/inria-00350638},
+  journal={{IEEE Robotics and Automation Magazine}},
+  publisher={{Institute of Electrical and Electronics Engineers}},
+  volume={14},
+  number={1},
+  pages={109-118},
+  year={2007},
+  pdf={https://inria.hal.science/inria-00350638/file/2007_ieee_ram_chaumette.pdf},
+  hal_id={inria-00350638},
+  hal_version={v1},
+}
+
+@article{Hutchinson1996ATO,
+  title={A tutorial on visual servo control},
+  author={Seth A. Hutchinson and Gregory Hager and Peter Corke},
+  journal={IEEE Trans. Robotics Autom.},
+  year={1996},
+  volume={12},
+  pages={651-670},
+  url={https://api.semanticscholar.org/CorpusID:1814423}
+}
diff --git a/modules/tracking/include/opencv2/tracking/twist.hpp b/modules/tracking/include/opencv2/tracking/twist.hpp
index 8d998beda33..1452a00cddd 100644
--- a/modules/tracking/include/opencv2/tracking/twist.hpp
+++ b/modules/tracking/include/opencv2/tracking/twist.hpp
@@ -16,7 +16,7 @@ inline namespace tracking
  * @brief Compute the camera twist from a set of 2D pixel locations, their
  * velocities, depth values and intrinsic parameters of the camera. The pixel
  * velocities are usually obtained from optical flow algorithms, both dense and
- * sparse flow can be used to compute the flow between images and duv computed by
+ * sparse flow can be used to compute the flow between images and \p duv computed by
  * dividing the flow by the time interval between the images.
  *
  * @param uv 2xN matrix of 2D pixel locations
@@ -30,9 +30,10 @@ CV_EXPORTS cv::Vec6d computeTwist(const cv::Mat& uv, const cv::Mat& duv, const c
                                   const cv::Mat& K);
 
 /**
- * @brief Compute the interaction matrix for a set of 2D pixels. This is usually
+ * @brief Compute the interaction matrix ( @cite Hutchinson1996ATO @cite chaumette:inria-00350283
+ * @cite chaumette:inria-00350638 ) for a set of 2D pixels. This is usually
  * used in visual servoing applications to command a robot to move at desired pixel
- * locations/velocities. By inverting this matrix one can estimate camera spatial
+ * locations/velocities. By inverting this matrix, one can estimate camera spatial
  * velocity i.e., the twist.
  *
  * @param uv 2xN matrix of 2D pixel locations
@@ -41,8 +42,8 @@ CV_EXPORTS cv::Vec6d computeTwist(const cv::Mat& uv, const cv::Mat& duv, const c
  * @param J 2Nx6 interaction matrix
  *
  */
-CV_EXPORTS void getInteractionMatrix(const cv::Mat& uv, const cv::Mat& depths, const cv::Mat& K,
-                                     cv::Mat& J);
+CV_EXPORTS void computeInteractionMatrix(const cv::Mat& uv, const cv::Mat& depths, const cv::Mat& K,
+                                         cv::Mat& J);
 
 //! @}
 
diff --git a/modules/tracking/src/twist.cpp b/modules/tracking/src/twist.cpp
index 1ff84c42582..bad6661ff1c 100644
--- a/modules/tracking/src/twist.cpp
+++ b/modules/tracking/src/twist.cpp
@@ -9,7 +9,7 @@ namespace detail
 inline namespace tracking
 {
 
-void getInteractionMatrix(const cv::Mat& uv, const cv::Mat& depths, const cv::Mat& K, cv::Mat& J)
+void computeInteractionMatrix(const cv::Mat& uv, const cv::Mat& depths, const cv::Mat& K, cv::Mat& J)
 {
     CV_Assert(uv.cols == depths.cols);
     CV_Assert(depths.type() == CV_32F);
@@ -64,7 +64,7 @@ cv::Vec6d computeTwist(const cv::Mat& uv, const cv::Mat& duv, const cv::Mat& dep
     CV_Assert(uv.cols * 2 == duv.rows);
 
     cv::Mat J;
-    getInteractionMatrix(uv, depths, K, J);
+    computeInteractionMatrix(uv, depths, K, J);
     cv::Mat Jinv;
     cv::invert(J, Jinv, cv::DECOMP_SVD);
     cv::Mat twist = Jinv * duv;
diff --git a/modules/tracking/test/test_twist.cpp b/modules/tracking/test/test_twist.cpp
index 3911f28aea8..f365d811848 100644
--- a/modules/tracking/test/test_twist.cpp
+++ b/modules/tracking/test/test_twist.cpp
@@ -39,7 +39,7 @@ TEST_F(TwistTest, TestInteractionMatrix)
     cv::Mat uv = cv::Mat(2, 1, CV_32F, {1.0f, 1.0f});
     cv::Mat depth = cv::Mat(1, 1, CV_32F, {2.0f});
 
-    getInteractionMatrix(uv, depth, K, J);
+    computeInteractionMatrix(uv, depth, K, J);
     ASSERT_EQ(J.cols, 6);
     ASSERT_EQ(J.rows, 2);
     float expected[2][6] = {{-0.5f, 0.0f, 0.5f, 1.0f, -2.0f, 1.0f},
@@ -87,7 +87,7 @@ TEST_F(TwistTest, TestComputeWithNonZeroPixelVelocities)
     float duv_data[] = {1.0f, 2.0f, 1.0f, 3.0f, 1.0f, 4.0f};
     cv::Mat duv = cv::Mat(6, 1, CV_32F, duv_data);
 
-    getInteractionMatrix(uv, depth, K, J);
+    computeInteractionMatrix(uv, depth, K, J);
     ASSERT_EQ(J.cols, 6);
     ASSERT_EQ(J.rows, 6);
     float expected_jac[6][6] = {{-1.0f, 0.0f, 1.0f, 1.0f, -2.0f, 1.0f},

From 4e766a039eda04b75f28f6905e4192d99b7874c7 Mon Sep 17 00:00:00 2001
From: sdy623 <syc@nbsyc.com>
Date: Tue, 23 Apr 2024 21:43:40 +0900
Subject: [PATCH 03/12] cudaarithm: fix the compile faiure of CUDA 12.4.x . A
 slight API change of NPP nppiMeanStdDevGetBufferHostSize_8u_C1R The type of
 bufSize is size_t instead of int in CUDA 12.4.x

---
 modules/cudaarithm/src/reductions.cpp | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/modules/cudaarithm/src/reductions.cpp b/modules/cudaarithm/src/reductions.cpp
index cfadad648a9..b70a128558f 100644
--- a/modules/cudaarithm/src/reductions.cpp
+++ b/modules/cudaarithm/src/reductions.cpp
@@ -151,7 +151,12 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, Stream& stream)
     sz.width  = gsrc.cols;
     sz.height = gsrc.rows;
 
+#if (CUDA_VERSION >= 12040)
+    size_t bufSize;
+#else
     int bufSize;
+#endif
+
 #if (CUDA_VERSION <= 4020)
     nppSafeCall( nppiMeanStdDev8uC1RGetBufferHostSize(sz, &bufSize) );
 #else
@@ -162,7 +167,8 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, Stream& stream)
 #endif
 
     BufferPool pool(stream);
-    GpuMat buf = pool.getBuffer(1, bufSize, gsrc.type());
+    CV_Assert(bufSize <= std::numeric_limits<int>::max());
+    GpuMat buf = pool.getBuffer(1, static_cast<int>(bufSize), gsrc.type());
 
     // detail: https://github.com/opencv/opencv/issues/11063
     //NppStreamHandler h(StreamAccessor::getStream(stream));
@@ -227,7 +233,12 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, InputArray mask, Stre
     sz.width  = gsrc.cols;
     sz.height = gsrc.rows;
 
+#if (CUDA_VERSION >= 12040)
+    size_t bufSize;
+#else
     int bufSize;
+#endif
+
 #if (CUDA_VERSION <= 4020)
         nppSafeCall( nppiMeanStdDev8uC1MRGetBufferHostSize(sz, &bufSize) );
 #else
@@ -238,7 +249,8 @@ void cv::cuda::meanStdDev(InputArray src, OutputArray dst, InputArray mask, Stre
 #endif
 
     BufferPool pool(stream);
-    GpuMat buf = pool.getBuffer(1, bufSize, gsrc.type());
+    CV_Assert(bufSize <= std::numeric_limits<int>::max());
+    GpuMat buf = pool.getBuffer(1, static_cast<int>(bufSize), gsrc.type());
 
     if(gsrc.type() == CV_8UC1)
         nppSafeCall( nppiMean_StdDev_8u_C1MR(gsrc.ptr<Npp8u>(), static_cast<int>(gsrc.step), gmask.ptr<Npp8u>(), static_cast<int>(gmask.step),

From 6b1faf0d9bc174c5fb40c4941bac615392436e81 Mon Sep 17 00:00:00 2001
From: Pavel Rojtberg <pavel.rojtberg@igd.fraunhofer.de>
Date: Wed, 15 May 2024 08:30:24 +0200
Subject: [PATCH 04/12] Merge pull request #3735 from paroj:ovisup

ovis: force camera extent update to get correct bbox #3735

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [ ] There is a reference to the original bug report and related work
- [ ] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [x] The feature is well documented and sample code can be built with the project CMake

---
fixes #3732
---
 modules/ovis/CMakeLists.txt            |  2 +-
 modules/ovis/samples/aruco_ar_demo.cpp | 14 ++++++++------
 modules/ovis/samples/aruco_ar_demo.py  |  2 +-
 modules/ovis/src/ovis.cpp              |  1 +
 4 files changed, 11 insertions(+), 8 deletions(-)

diff --git a/modules/ovis/CMakeLists.txt b/modules/ovis/CMakeLists.txt
index 912b86a590d..aff20c2b63e 100644
--- a/modules/ovis/CMakeLists.txt
+++ b/modules/ovis/CMakeLists.txt
@@ -24,7 +24,7 @@ ocv_glob_module_sources()
 ocv_module_include_directories()
 ocv_create_module()
 
-ocv_add_samples(opencv_aruco)
+ocv_add_samples(opencv_objdetect opencv_aruco)
 
 ocv_warnings_disable(CMAKE_CXX_FLAGS -Wunused-parameter)
 ocv_target_link_libraries(${the_module} ${OGRE_LIBRARIES})
diff --git a/modules/ovis/samples/aruco_ar_demo.cpp b/modules/ovis/samples/aruco_ar_demo.cpp
index 2398a7182a3..0b265beea84 100644
--- a/modules/ovis/samples/aruco_ar_demo.cpp
+++ b/modules/ovis/samples/aruco_ar_demo.cpp
@@ -3,7 +3,7 @@
 #include <opencv2/videoio.hpp>
 
 #include <opencv2/ovis.hpp>
-#include <opencv2/aruco_detector.hpp>
+#include <opencv2/aruco.hpp>
 
 #include <iostream>
 
@@ -24,10 +24,12 @@ int main()
   const double focal_length = 800.0;
 
   // aruco
-  Ptr<aruco::Dictionary> adict = aruco::getPredefinedDictionary(aruco::DICT_4X4_50);
-  //Mat out_img;
-  //aruco::drawMarker(adict, 0, 400, out_img);
-  //imshow("marker", out_img);
+  aruco::Dictionary adict = aruco::getPredefinedDictionary(aruco::DICT_4X4_50);
+
+  aruco::ArucoDetector detector(adict);
+  Mat out_img;
+  adict.generateImageMarker(0, 400, out_img);
+  imshow("marker", out_img);
 
   // random calibration data, your mileage may vary
   Mat1d cm = Mat1d::zeros(3, 3); // init empty matrix
@@ -53,7 +55,7 @@ int main()
   while (ovis::waitKey(1) != KEY_ESCAPE) {
     cap.read(img);
     win->setBackground(img);
-    aruco::detectMarkers(img, adict, corners, ids);
+    detector.detectMarkers(img, corners, ids);
 
     waitKey(1);
 
diff --git a/modules/ovis/samples/aruco_ar_demo.py b/modules/ovis/samples/aruco_ar_demo.py
index 72aeeaebea3..5877d915d9e 100644
--- a/modules/ovis/samples/aruco_ar_demo.py
+++ b/modules/ovis/samples/aruco_ar_demo.py
@@ -3,7 +3,7 @@
 
 # aruco
 adict = cv.aruco.getPredefinedDictionary(cv.aruco.DICT_4X4_50)
-cv.imshow("marker", cv.aruco.drawMarker(adict, 0, 400))
+cv.imshow("marker", adict.generateImageMarker(0, 400))
 
 # random calibration data. your mileage may vary.
 imsize = (800, 600)
diff --git a/modules/ovis/src/ovis.cpp b/modules/ovis/src/ovis.cpp
index 8c0e3431e4c..38dfcbc2ee0 100644
--- a/modules/ovis/src/ovis.cpp
+++ b/modules/ovis/src/ovis.cpp
@@ -588,6 +588,7 @@ class WindowSceneImpl : public WindowScene
         cam->setDebugDisplayEnabled(true);
         cam->setNearClipDistance(1e-9);
         cam->setFarClipDistance(zFar);
+        cam->getFrustumExtents(); // force update
 
         _setCameraIntrinsics(cam, K, imsize);
 

From cb08ac6b2ea890e3355c27caa782d3bfb1a7da69 Mon Sep 17 00:00:00 2001
From: Alexander Smorkalov <2536374+asmorkalov@users.noreply.github.com>
Date: Wed, 15 May 2024 14:20:55 +0300
Subject: [PATCH 05/12] Merge pull request #3734 from
 asmorkalov:as/std_move_warning

Fixed Wredundant-move produced by GCC 13.2 (Ubuntu 24.04). #3734

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [x] There is a reference to the original bug report and related work
- [ ] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [ ] The feature is well documented and sample code can be built with the project CMake
---
 modules/face/src/facemarkLBF.cpp              | 32 ++++---
 modules/face/src/mace.cpp                     |  4 +-
 modules/mcc/test/test_mcc.cpp                 |  6 +-
 modules/rgbd/perf/perf_tsdf.cpp               |  6 +-
 modules/rgbd/test/ocl/test_tsdf.cpp           |  2 +-
 modules/rgbd/test/test_colored_kinfu.cpp      | 20 ++---
 modules/rgbd/test/test_kinfu.cpp              |  4 +-
 modules/rgbd/test/test_tsdf.cpp               |  2 +-
 modules/videostab/src/global_motion.cpp       | 83 ++++++++++---------
 .../test/test_sparse_match_interpolator.cpp   | 14 ++--
 10 files changed, 87 insertions(+), 86 deletions(-)

diff --git a/modules/face/src/facemarkLBF.cpp b/modules/face/src/facemarkLBF.cpp
index df58c6319b7..49cd8f63fea 100644
--- a/modules/face/src/facemarkLBF.cpp
+++ b/modules/face/src/facemarkLBF.cpp
@@ -661,24 +661,22 @@ FacemarkLBFImpl::BBox::BBox(double _x, double _y, double w, double h) {
 
 // Project absolute shape to relative shape binding to this bbox
 Mat FacemarkLBFImpl::BBox::project(const Mat &shape) const {
-    Mat_<double> res(shape.rows, shape.cols);
-    const Mat_<double> &shape_ = (Mat_<double>)shape;
+    Mat res(shape.rows, shape.cols, CV_64FC1);
     for (int i = 0; i < shape.rows; i++) {
-        res(i, 0) = (shape_(i, 0) - x_center) / x_scale;
-        res(i, 1) = (shape_(i, 1) - y_center) / y_scale;
+        res.at<double>(i, 0) = (shape.at<double>(i, 0) - x_center) / x_scale;
+        res.at<double>(i, 1) = (shape.at<double>(i, 1) - y_center) / y_scale;
     }
-    return std::move(res);
+    return res;
 }
 
 // Project relative shape to absolute shape binding to this bbox
 Mat FacemarkLBFImpl::BBox::reproject(const Mat &shape) const {
-    Mat_<double> res(shape.rows, shape.cols);
-    const Mat_<double> &shape_ = (Mat_<double>)shape;
+    Mat res(shape.rows, shape.cols, CV_64FC1);
     for (int i = 0; i < shape.rows; i++) {
-        res(i, 0) = shape_(i, 0)*x_scale + x_center;
-        res(i, 1) = shape_(i, 1)*y_scale + y_center;
+        res.at<double>(i, 0) = shape.at<double>(i, 0)*x_scale + x_center;
+        res.at<double>(i, 1) = shape.at<double>(i, 1)*y_scale + y_center;
     }
-    return std::move(res);
+    return res;
 }
 
 Mat FacemarkLBFImpl::getMeanShape(std::vector<Mat> &gt_shapes, std::vector<BBox> &bboxes) {
@@ -997,7 +995,7 @@ void FacemarkLBFImpl::RandomForest::train(std::vector<Mat> &imgs, std::vector<Ma
 }
 
 Mat FacemarkLBFImpl::RandomForest::generateLBF(Mat &img, Mat &current_shape, BBox &bbox, Mat &mean_shape) {
-    Mat_<int> lbf_feat(1, landmark_n*trees_n);
+    Mat lbf_feat(1, landmark_n*trees_n, CV_32SC1);
     double scale;
     Mat_<double> rotate;
     calcSimilarityTransform(bbox.project(current_shape), mean_shape, scale, rotate);
@@ -1036,10 +1034,10 @@ Mat FacemarkLBFImpl::RandomForest::generateLBF(Mat &img, Mat &current_shape, BBo
                     idx = 2 * idx + 1;
                 }
             }
-            lbf_feat(i*trees_n + j) = (i*trees_n + j)*base + code;
+            lbf_feat.at<int>(i*trees_n + j) = (i*trees_n + j)*base + code;
         }
     }
-    return std::move(lbf_feat);
+    return lbf_feat;
 }
 
 void FacemarkLBFImpl::RandomForest::write(FileStorage fs, int k) {
@@ -1365,7 +1363,7 @@ Mat FacemarkLBFImpl::Regressor::supportVectorRegression(
 
 Mat FacemarkLBFImpl::Regressor::globalRegressionPredict(const Mat &lbf, int stage) {
     const Mat_<double> &weight = (Mat_<double>)gl_regression_weights[stage];
-    Mat_<double> delta_shape(weight.rows / 2, 2);
+    Mat delta_shape(weight.rows / 2, 2, CV_64FC1);
     const double *w_ptr = NULL;
     const int *lbf_ptr = lbf.ptr<int>(0);
 
@@ -1374,14 +1372,14 @@ Mat FacemarkLBFImpl::Regressor::globalRegressionPredict(const Mat &lbf, int stag
         w_ptr = weight.ptr<double>(2 * i);
         double y = 0;
         for (int j = 0; j < lbf.cols; j++) y += w_ptr[lbf_ptr[j]];
-        delta_shape(i, 0) = y;
+        delta_shape.at<double>(i, 0) = y;
 
         w_ptr = weight.ptr<double>(2 * i + 1);
         y = 0;
         for (int j = 0; j < lbf.cols; j++) y += w_ptr[lbf_ptr[j]];
-        delta_shape(i, 1) = y;
+        delta_shape.at<double>(i, 1) = y;
     }
-    return std::move(delta_shape);
+    return delta_shape;
 } // Regressor::globalRegressionPredict
 
 Mat FacemarkLBFImpl::Regressor::predict(Mat &img, BBox &bbox) {
diff --git a/modules/face/src/mace.cpp b/modules/face/src/mace.cpp
index 2c09560bfe5..3b7a236828c 100644
--- a/modules/face/src/mace.cpp
+++ b/modules/face/src/mace.cpp
@@ -102,11 +102,11 @@ struct MACEImpl CV_FINAL : MACE {
         Mat complexInput;
         merge(input, 2, complexInput);
 
-        Mat_<Vec2d> dftImg(IMGSIZE*2, IMGSIZE*2, 0.0);
+        Mat dftImg(IMGSIZE*2, IMGSIZE*2, CV_64FC2, 0.0);
         complexInput.copyTo(dftImg(Rect(0,0,IMGSIZE,IMGSIZE)));
 
         dft(dftImg, dftImg);
-        return std::move(dftImg);
+        return dftImg;
     }
 
 
diff --git a/modules/mcc/test/test_mcc.cpp b/modules/mcc/test/test_mcc.cpp
index 374b829b4b2..37bd1f11fc7 100644
--- a/modules/mcc/test/test_mcc.cpp
+++ b/modules/mcc/test/test_mcc.cpp
@@ -102,7 +102,9 @@ TEST(CV_mcc_ccm_test, detect_Macbeth)
 
     // check Macbeth corners
     vector<Point2f> corners = checker->getBox();
-    EXPECT_MAT_NEAR(gold_corners, corners, 3.6); // diff 3.57385 in ARM only
+    // diff 3.57385 corresponds to ARM v8
+    // diff 4.37915 correspnds to Ubuntu 24.04 x86_64 configuration
+    EXPECT_MAT_NEAR(gold_corners, corners, 4.38);
 
     // read gold chartsRGB
     node = fs["chartsRGB"];
@@ -112,7 +114,7 @@ TEST(CV_mcc_ccm_test, detect_Macbeth)
 
     // check chartsRGB
     Mat chartsRGB = checker->getChartsRGB();
-    EXPECT_MAT_NEAR(goldChartsRGB.col(1), chartsRGB.col(1), 0.25); // diff 0.240634 in ARM only
+    EXPECT_MAT_NEAR(goldChartsRGB.col(1), chartsRGB.col(1), 0.3); // diff 0.292077 on Ubuntu 20.04 ARM64
 }
 
 TEST(CV_mcc_ccm_test, compute_ccm)
diff --git a/modules/rgbd/perf/perf_tsdf.cpp b/modules/rgbd/perf/perf_tsdf.cpp
index da928f98f22..999ddb51673 100644
--- a/modules/rgbd/perf/perf_tsdf.cpp
+++ b/modules/rgbd/perf/perf_tsdf.cpp
@@ -91,7 +91,7 @@ struct Scene
 {
     virtual ~Scene() {}
     static Ptr<Scene> create(Size sz, Matx33f _intr, float _depthFactor, bool onlySemisphere);
-    virtual Mat depth(Affine3f pose) = 0;
+    virtual Mat_<float> depth(const Affine3f& pose) = 0;
     virtual std::vector<Affine3f> getPoses() = 0;
 };
 
@@ -131,7 +131,7 @@ struct SemisphereScene : Scene
         return res;
     }
 
-    Mat depth(Affine3f pose) override
+    Mat_<float> depth(const Affine3f& pose) override
     {
         Mat_<float> frame(frameSize);
         Reprojector reproj(intr);
@@ -139,7 +139,7 @@ struct SemisphereScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<SemisphereScene>(frame, pose, reproj, depthFactor, onlySemisphere));
 
-        return std::move(frame);
+        return frame;
     }
 
     std::vector<Affine3f> getPoses() override
diff --git a/modules/rgbd/test/ocl/test_tsdf.cpp b/modules/rgbd/test/ocl/test_tsdf.cpp
index 1c55e1f4001..fa3d593cb4f 100644
--- a/modules/rgbd/test/ocl/test_tsdf.cpp
+++ b/modules/rgbd/test/ocl/test_tsdf.cpp
@@ -143,7 +143,7 @@ struct SemisphereScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<SemisphereScene>(frame, pose, reproj, depthFactor, onlySemisphere));
 
-        return std::move(frame);
+        return static_cast<Mat>(frame);
     }
 
     std::vector<Affine3f> getPoses() override
diff --git a/modules/rgbd/test/test_colored_kinfu.cpp b/modules/rgbd/test/test_colored_kinfu.cpp
index 4303b260e3e..0ff90704e93 100644
--- a/modules/rgbd/test/test_colored_kinfu.cpp
+++ b/modules/rgbd/test/test_colored_kinfu.cpp
@@ -158,8 +158,8 @@ struct Scene
 {
     virtual ~Scene() {}
     static Ptr<Scene> create(int nScene, Size sz, Matx33f _intr, float _depthFactor);
-    virtual Mat depth(Affine3f pose) = 0;
-    virtual Mat rgb(Affine3f pose) = 0;
+    virtual Mat_<float> depth(const Affine3f& pose) = 0;
+    virtual Mat_<Vec3f> rgb(const Affine3f& pose) = 0;
     virtual std::vector<Affine3f> getPoses() = 0;
 };
 
@@ -198,7 +198,7 @@ struct CubeSpheresScene : Scene
         return res;
     }
 
-    Mat depth(Affine3f pose) override
+    Mat_<float> depth(const Affine3f& pose) override
     {
         Mat_<float> frame(frameSize);
         Reprojector reproj(intr);
@@ -206,10 +206,10 @@ struct CubeSpheresScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<CubeSpheresScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return frame;
     }
 
-    Mat rgb(Affine3f pose) override
+    Mat_<Vec3f> rgb(const Affine3f& pose) override
     {
         Mat_<Vec3f> frame(frameSize);
         Reprojector reproj(intr);
@@ -217,7 +217,7 @@ struct CubeSpheresScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderColorInvoker<CubeSpheresScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return frame;
     }
 
     std::vector<Affine3f> getPoses() override
@@ -305,7 +305,7 @@ struct RotatingScene : Scene
         return res;
     }
 
-    Mat depth(Affine3f pose) override
+    Mat_<float> depth(const Affine3f& pose) override
     {
         Mat_<float> frame(frameSize);
         Reprojector reproj(intr);
@@ -313,10 +313,10 @@ struct RotatingScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<RotatingScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return frame;
     }
 
-    Mat rgb(Affine3f pose) override
+    Mat_<Vec3f> rgb(const Affine3f& pose) override
     {
         Mat_<Vec3f> frame(frameSize);
         Reprojector reproj(intr);
@@ -324,7 +324,7 @@ struct RotatingScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderColorInvoker<RotatingScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return frame;
     }
 
     std::vector<Affine3f> getPoses() override
diff --git a/modules/rgbd/test/test_kinfu.cpp b/modules/rgbd/test/test_kinfu.cpp
index 18059013ee7..e9c821f1577 100644
--- a/modules/rgbd/test/test_kinfu.cpp
+++ b/modules/rgbd/test/test_kinfu.cpp
@@ -141,7 +141,7 @@ struct CubeSpheresScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<CubeSpheresScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return static_cast<Mat>(frame);
     }
 
     std::vector<Affine3f> getPoses() override
@@ -237,7 +237,7 @@ struct RotatingScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<RotatingScene>(frame, pose, reproj, depthFactor));
 
-        return std::move(frame);
+        return static_cast<Mat>(frame);
     }
 
     std::vector<Affine3f> getPoses() override
diff --git a/modules/rgbd/test/test_tsdf.cpp b/modules/rgbd/test/test_tsdf.cpp
index 31a137854c3..f5a88180923 100644
--- a/modules/rgbd/test/test_tsdf.cpp
+++ b/modules/rgbd/test/test_tsdf.cpp
@@ -140,7 +140,7 @@ struct SemisphereScene : Scene
         Range range(0, frame.rows);
         parallel_for_(range, RenderInvoker<SemisphereScene>(frame, pose, reproj, depthFactor, onlySemisphere));
 
-        return std::move(frame);
+        return static_cast<Mat>(frame);
     }
 
     std::vector<Affine3f> getPoses() override
diff --git a/modules/videostab/src/global_motion.cpp b/modules/videostab/src/global_motion.cpp
index 5bef5b1e74f..eb42d6954d6 100644
--- a/modules/videostab/src/global_motion.cpp
+++ b/modules/videostab/src/global_motion.cpp
@@ -82,7 +82,7 @@ namespace videostab
 {
 
 // does isotropic normalization
-static Mat normalizePoints(int npoints, Point2f *points)
+static Mat_<float> normalizePoints(int npoints, Point2f *points)
 {
     float cx = 0.f, cy = 0.f;
     for (int i = 0; i < npoints; ++i)
@@ -113,32 +113,32 @@ static Mat normalizePoints(int npoints, Point2f *points)
     T(0,0) = T(1,1) = s;
     T(0,2) = -cx*s;
     T(1,2) = -cy*s;
-    return std::move(T);
+    return T;
 }
 
 
 static Mat estimateGlobMotionLeastSquaresTranslation(
         int npoints, Point2f *points0, Point2f *points1, float *rmse)
 {
-    Mat_<float> M = Mat::eye(3, 3, CV_32F);
+    Mat M = Mat::eye(3, 3, CV_32FC1);
     for (int i = 0; i < npoints; ++i)
     {
-        M(0,2) += points1[i].x - points0[i].x;
-        M(1,2) += points1[i].y - points0[i].y;
+        M.at<float>(0,2) += points1[i].x - points0[i].x;
+        M.at<float>(1,2) += points1[i].y - points0[i].y;
     }
-    M(0,2) /= npoints;
-    M(1,2) /= npoints;
+    M.at<float>(0,2) /= npoints;
+    M.at<float>(1,2) /= npoints;
 
     if (rmse)
     {
         *rmse = 0;
         for (int i = 0; i < npoints; ++i)
-            *rmse += sqr(points1[i].x - points0[i].x - M(0,2)) +
-                     sqr(points1[i].y - points0[i].y - M(1,2));
+            *rmse += sqr(points1[i].x - points0[i].x - M.at<float>(0,2)) +
+                     sqr(points1[i].y - points0[i].y - M.at<float>(1,2));
         *rmse = std::sqrt(*rmse / npoints);
     }
 
-    return std::move(M);
+    return M;
 }
 
 
@@ -194,16 +194,16 @@ static Mat estimateGlobMotionLeastSquaresRotation(
 
     // A*sin(alpha) + B*cos(alpha) = 0
     float C = std::sqrt(A*A + B*B);
-    Mat_<float> M = Mat::eye(3, 3, CV_32F);
+    Mat M = Mat::eye(3, 3, CV_32F);
     if ( C != 0 )
     {
         float sinAlpha = - B / C;
         float cosAlpha = A / C;
 
-        M(0,0) = cosAlpha;
-        M(1,1) = M(0,0);
-        M(0,1) = sinAlpha;
-        M(1,0) = - M(0,1);
+        M.at<float>(0,0) = cosAlpha;
+        M.at<float>(1,1) = cosAlpha;
+        M.at<float>(0,1) = sinAlpha;
+        M.at<float>(1,0) = - sinAlpha;
     }
 
     if (rmse)
@@ -213,16 +213,16 @@ static Mat estimateGlobMotionLeastSquaresRotation(
         {
             p0 = points0[i];
             p1 = points1[i];
-            *rmse += sqr(p1.x - M(0,0)*p0.x - M(0,1)*p0.y) +
-                     sqr(p1.y - M(1,0)*p0.x - M(1,1)*p0.y);
+            *rmse += sqr(p1.x - M.at<float>(0,0)*p0.x - M.at<float>(0,1)*p0.y) +
+                     sqr(p1.y - M.at<float>(1,0)*p0.x - M.at<float>(1,1)*p0.y);
         }
         *rmse = std::sqrt(*rmse / npoints);
     }
 
-    return std::move(M);
+    return M;
 }
 
-static Mat  estimateGlobMotionLeastSquaresRigid(
+static Mat estimateGlobMotionLeastSquaresRigid(
         int npoints, Point2f *points0, Point2f *points1, float *rmse)
 {
     Point2f mean0(0.f, 0.f);
@@ -250,15 +250,15 @@ static Mat  estimateGlobMotionLeastSquaresRigid(
         A(1,1) += pt1.y * pt0.y;
     }
 
-    Mat_<float> M = Mat::eye(3, 3, CV_32F);
+    Mat M = Mat::eye(3, 3, CV_32FC1);
 
     SVD svd(A);
     Mat_<float> R = svd.u * svd.vt;
     Mat tmp(M(Rect(0,0,2,2)));
     R.copyTo(tmp);
 
-    M(0,2) = mean1.x - R(0,0)*mean0.x - R(0,1)*mean0.y;
-    M(1,2) = mean1.y - R(1,0)*mean0.x - R(1,1)*mean0.y;
+    M.at<float>(0,2) = mean1.x - R(0,0)*mean0.x - R(0,1)*mean0.y;
+    M.at<float>(1,2) = mean1.y - R(1,0)*mean0.x - R(1,1)*mean0.y;
 
     if (rmse)
     {
@@ -267,13 +267,13 @@ static Mat  estimateGlobMotionLeastSquaresRigid(
         {
             pt0 = points0[i];
             pt1 = points1[i];
-            *rmse += sqr(pt1.x - M(0,0)*pt0.x - M(0,1)*pt0.y - M(0,2)) +
-                     sqr(pt1.y - M(1,0)*pt0.x - M(1,1)*pt0.y - M(1,2));
+            *rmse += sqr(pt1.x - M.at<float>(0,0)*pt0.x - M.at<float>(0,1)*pt0.y - M.at<float>(0,2)) +
+                     sqr(pt1.y - M.at<float>(1,0)*pt0.x - M.at<float>(1,1)*pt0.y - M.at<float>(1,2));
         }
         *rmse = std::sqrt(*rmse / npoints);
     }
 
-    return std::move(M);
+    return M;
 }
 
 
@@ -404,7 +404,7 @@ Mat estimateGlobalMotionRansac(
     // best hypothesis
     std::vector<int> bestIndices(params.size);
 
-    Mat_<float> bestM;
+    Mat bestM;
     int ninliersMax = -1;
 
     RNG rng(0);
@@ -469,8 +469,8 @@ Mat estimateGlobalMotionRansac(
         {
             p0 = points0_[i];
             p1 = points1_[i];
-            x = bestM(0,0)*p0.x + bestM(0,1)*p0.y + bestM(0,2);
-            y = bestM(1,0)*p0.x + bestM(1,1)*p0.y + bestM(1,2);
+            x = bestM.at<float>(0,0)*p0.x + bestM.at<float>(0,1)*p0.y + bestM.at<float>(0,2);
+            y = bestM.at<float>(1,0)*p0.x + bestM.at<float>(1,1)*p0.y + bestM.at<float>(1,2);
             if (sqr(x - p1.x) + sqr(y - p1.y) < params.thresh * params.thresh)
             {
                 subset0[j] = p0;
@@ -484,7 +484,7 @@ Mat estimateGlobalMotionRansac(
     if (ninliers)
         *ninliers = ninliersMax;
 
-    return std::move(bestM);
+    return bestM;
 }
 
 
@@ -505,7 +505,7 @@ Mat MotionEstimatorRansacL2::estimate(InputArray points0, InputArray points1, bo
     // find motion
 
     int ninliers = 0;
-    Mat_<float> M;
+    Mat M;
 
     if (motionModel() != MM_HOMOGRAPHY)
         M = estimateGlobalMotionRansac(
@@ -527,7 +527,7 @@ Mat MotionEstimatorRansacL2::estimate(InputArray points0, InputArray points1, bo
         if (ok) *ok = false;
     }
 
-    return std::move(M);
+    return M;
 }
 
 
@@ -675,13 +675,13 @@ FromFileMotionReader::FromFileMotionReader(const String &path)
 
 Mat FromFileMotionReader::estimate(const Mat &/*frame0*/, const Mat &/*frame1*/, bool *ok)
 {
-    Mat_<float> M(3, 3);
+    Mat M(3, 3, CV_32FC1);
     bool ok_;
-    file_ >> M(0,0) >> M(0,1) >> M(0,2)
-          >> M(1,0) >> M(1,1) >> M(1,2)
-          >> M(2,0) >> M(2,1) >> M(2,2) >> ok_;
+    file_ >> M.at<float>(0,0) >> M.at<float>(0,1) >> M.at<float>(0,2)
+          >> M.at<float>(1,0) >> M.at<float>(1,1) >> M.at<float>(1,2)
+          >> M.at<float>(2,0) >> M.at<float>(2,1) >> M.at<float>(2,2) >> ok_;
     if (ok) *ok = ok_;
-    return std::move(M);
+    return M;
 }
 
 
@@ -696,12 +696,13 @@ ToFileMotionWriter::ToFileMotionWriter(const String &path, Ptr<ImageMotionEstima
 Mat ToFileMotionWriter::estimate(const Mat &frame0, const Mat &frame1, bool *ok)
 {
     bool ok_;
-    Mat_<float> M = motionEstimator_->estimate(frame0, frame1, &ok_);
-    file_ << M(0,0) << " " << M(0,1) << " " << M(0,2) << " "
-          << M(1,0) << " " << M(1,1) << " " << M(1,2) << " "
-          << M(2,0) << " " << M(2,1) << " " << M(2,2) << " " << ok_ << std::endl;
+    Mat M = motionEstimator_->estimate(frame0, frame1, &ok_);
+    file_ << M.at<float>(0,0) << " " << M.at<float>(0,1) << " " << M.at<float>(0,2) << " "
+          << M.at<float>(1,0) << " " << M.at<float>(1,1) << " " << M.at<float>(1,2) << " "
+          << M.at<float>(2,0) << " " << M.at<float>(2,1) << " " << M.at<float>(2,2) << " "
+          << ok_ << std::endl;
     if (ok) *ok = ok_;
-    return std::move(M);
+    return M;
 }
 
 
diff --git a/modules/ximgproc/test/test_sparse_match_interpolator.cpp b/modules/ximgproc/test/test_sparse_match_interpolator.cpp
index 261d6109bf9..f53a5add027 100644
--- a/modules/ximgproc/test/test_sparse_match_interpolator.cpp
+++ b/modules/ximgproc/test/test_sparse_match_interpolator.cpp
@@ -17,22 +17,22 @@ Mat readOpticalFlow( const String& path )
 //    CV_Assert(sizeof(float) == 4);
     //FIXME: ensure right sizes of int and float - here and in writeOpticalFlow()
 
-    Mat_<Point2f> flow;
+    Mat flow;
     std::ifstream file(path.c_str(), std::ios_base::binary);
     if ( !file.good() )
-        return std::move(flow); // no file - return empty matrix
+        return flow; // no file - return empty matrix
 
     float tag;
     file.read((char*) &tag, sizeof(float));
     if ( tag != FLOW_TAG_FLOAT )
-        return std::move(flow);
+        return flow;
 
     int width, height;
 
     file.read((char*) &width, 4);
     file.read((char*) &height, 4);
 
-    flow.create(height, width);
+    flow.create(height, width, CV_32FC2);
 
     for ( int i = 0; i < flow.rows; ++i )
     {
@@ -44,14 +44,14 @@ Mat readOpticalFlow( const String& path )
             if ( !file.good() )
             {
                 flow.release();
-                return std::move(flow);
+                return flow;
             }
 
-            flow(i, j) = u;
+            flow.at<Point2f>(i, j) = u;
         }
     }
     file.close();
-    return std::move(flow);
+    return flow;
 }
 
 CV_ENUM(GuideTypes, CV_8UC1, CV_8UC3)

From 96c9c85d654aaaccbf9542a4450a1396712caa83 Mon Sep 17 00:00:00 2001
From: Alexander Smorkalov <alexander.smorkalov@xperience.ai>
Date: Wed, 15 May 2024 17:03:40 +0300
Subject: [PATCH 06/12] More warning fixes for Ubuntu 24.04.

---
 modules/intensity_transform/src/bimef.cpp | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/modules/intensity_transform/src/bimef.cpp b/modules/intensity_transform/src/bimef.cpp
index 2500d2abe8a..9cad67578f0 100644
--- a/modules/intensity_transform/src/bimef.cpp
+++ b/modules/intensity_transform/src/bimef.cpp
@@ -226,15 +226,15 @@ static Mat solveLinearEquation(const Mat_<float>& img, Mat_<float>& W_h_, Mat_<f
     Eigen::Map<const Eigen::VectorXf> tin(img_t.ptr<float>(), img_t.rows*img_t.cols);
     Eigen::VectorXf x = cg.solve(tin);
 
-    Mat_<float> tout(img.rows, img.cols);
-    tout.forEach(
+    Mat tout(img.rows, img.cols, CV_32FC1);
+    tout.forEach<float>(
         [&](float &pixel, const int * position) -> void
         {
             pixel = x(position[1]*img.rows + position[0]);
         }
     );
 
-    return std::move(tout);
+    return tout;
 }
 
 static Mat_<float> tsmooth(const Mat_<float>& src, float lambda=0.01f, float sigma=3.0f, float sharpness=0.001f)

From ff694dbc901adadfb708fa809f9f44225ebdd3c3 Mon Sep 17 00:00:00 2001
From: Alexander Smorkalov <alexander.smorkalov@xperience.ai>
Date: Thu, 16 May 2024 14:59:02 +0300
Subject: [PATCH 07/12] Added Ubuntu 24.04 to regular CI.

---
 .github/workflows/PR-4.x.yaml | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/.github/workflows/PR-4.x.yaml b/.github/workflows/PR-4.x.yaml
index f33cc37d5d5..586d406b209 100644
--- a/.github/workflows/PR-4.x.yaml
+++ b/.github/workflows/PR-4.x.yaml
@@ -15,6 +15,9 @@ jobs:
   Ubuntu2204-x64:
     uses: opencv/ci-gha-workflow/.github/workflows/OCV-Contrib-PR-4.x-U22.yaml@main
 
+  Ubuntu2404-x64:
+    uses: opencv/ci-gha-workflow/.github/workflows/OCV-Contrib-PR-4.x-U24.yaml@main
+
   Ubuntu2004-x64-CUDA:
     uses: opencv/ci-gha-workflow/.github/workflows/OCV-Contrib-PR-4.x-U20-Cuda.yaml@main
 

From e46ba342b251fdbe0df34be0da1ba85577aa1e94 Mon Sep 17 00:00:00 2001
From: TumoiYorozu <morojuyi@gmail.com>
Date: Wed, 22 May 2024 17:53:03 +0900
Subject: [PATCH 08/12] Merge pull request #3627 from
 TumoiYorozu:wavelet_matrix_median_filter_cuda
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

 Implemented fast median filter for CUDA using Wavelet Matrix, a constant-time, HDR-compatible method #3627

I replaced the existing CUDA implementation of the histogram-based median filter with an implementation of a new wavelet matrix-based median filter algorithm, which I presented at SIGGRAPH Asia 2022.
This paper won the Best Paper Award in the journal track of technical papers (ACM Transactions on Graphics).

This new algorithm, like the histogram method, has the property that the window radius does not affect the computation time, and is several times faster than the histogram method. Furthermore, while the histogram method does not support HDR and only supports 8U images, this new algorithm supports HDR and also supports 16U and 32F images.

I (the author) have published the implementation on my personal GitHub and made some modifications for OpenCV to make it accessible from OpenCV. I used the CUB library, which is part of the standard toolkit since CUDA 11.0. Therefore, depending on the CUDA_VERSION, the code is written to use the new algorithm for versions 11.0 and above, and the existing histogram method for versions 10 and below.

Regarding the old histogram-based code, the CPU version of the median filter supports 16U and 32F for window sizes up to 5, but it seems that the histogram CUDA version of the median filter does not. Also, the number of channels supported is different: the CPU version supports 1, 3, and 4 channels, while the CUDA version supports only 1 channel. In addition, for the CUDA version of the histogram method, pixels at the edges of the image, i.e. where the window is insufficient, were set to zero. For example, if the window size is 7, the width of the 3 pixels at the top, bottom, left, and right were not calculated correctly. When checking the tests, it was found that they compared with the CPU version by cropping the edges with rect, and also the cropping area was too wide, with 8 pixels cropped from the top, bottom, left, and right when the window size was 7.

In this PR, I first corrected the rect range for the tests so that both the old histogram method and the new wavelet matrix method can pass. Also, the CUDA version now supports 16U, 32F, and multi-channel formats such as 3 and 4 channels. In addition, while the CPU version only supports window sizes up to 5 for HDR, the new CUDA Wavelet Matrix method supports sizes of 7 and above. Additionally, I have added new tests for 16U, 32F, and multi-channel formats, specifically 3 and 4 channels.

Paper’s project page: [Constant Time Median Filter using 2D Wavelet Matrix | Interactive Graphics & Engineering Lab](https://cgenglab.github.io/en/publication/sigga22_wmatrix_median/)
My implementation (as author): [GitHub - TumoiYorozu/WMatrixMedian](https://github.com/TumoiYorozu/WMatrixMedian)

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
~~- [ ] There is a reference to the original bug report and related work~~
- [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [x] The feature is well documented and sample code can be built with the project CMake
---
 modules/cudafilters/src/cuda/median_filter.cu |   79 ++
 .../src/cuda/wavelet_matrix_2d.cuh            | 1053 +++++++++++++++++
 .../wavelet_matrix_feature_support_checks.h   |   82 ++
 .../cuda/wavelet_matrix_float_supporter.cuh   |  227 ++++
 .../src/cuda/wavelet_matrix_multi.cuh         |  636 ++++++++++
 modules/cudafilters/src/filtering.cpp         |   33 +-
 modules/cudafilters/test/test_filters.cpp     |   18 +-
 7 files changed, 2124 insertions(+), 4 deletions(-)
 create mode 100644 modules/cudafilters/src/cuda/wavelet_matrix_2d.cuh
 create mode 100644 modules/cudafilters/src/cuda/wavelet_matrix_feature_support_checks.h
 create mode 100644 modules/cudafilters/src/cuda/wavelet_matrix_float_supporter.cuh
 create mode 100644 modules/cudafilters/src/cuda/wavelet_matrix_multi.cuh

diff --git a/modules/cudafilters/src/cuda/median_filter.cu b/modules/cudafilters/src/cuda/median_filter.cu
index ed46eb4bf94..6776428ae1a 100644
--- a/modules/cudafilters/src/cuda/median_filter.cu
+++ b/modules/cudafilters/src/cuda/median_filter.cu
@@ -53,6 +53,17 @@
 #include "opencv2/core/cuda/saturate_cast.hpp"
 #include "opencv2/core/cuda/border_interpolate.hpp"
 
+
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#include "wavelet_matrix_feature_support_checks.h"
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+#include "wavelet_matrix_multi.cuh"
+#include "wavelet_matrix_2d.cuh"
+#include "wavelet_matrix_float_supporter.cuh"
+#endif
+
+
 namespace cv { namespace cuda { namespace device
 {
         __device__ void histogramAddAndSub8(int* H, const int * hist_colAdd,const int * hist_colSub){
@@ -334,4 +345,72 @@ namespace cv { namespace cuda { namespace device
 
 }}}
 
+
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+namespace cv { namespace cuda { namespace device
+    {
+        using namespace wavelet_matrix_median;
+
+        template<int CH_NUM, typename T>
+        void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<T> src, PtrStepSz<T> dst, int radius,cudaStream_t stream){
+
+            constexpr bool is_float = std::is_same<T, float>::value;
+            constexpr static int WORD_SIZE = 32;
+            constexpr static int ThW = (std::is_same<T, uint8_t>::value ?  8 : 4);
+            constexpr static int ThH = (std::is_same<T, uint8_t>::value ? 64 : 256);
+            using XYIdxT = uint32_t;
+            using XIdxT = uint16_t;
+            using WM_T = typename std::conditional<is_float, uint32_t, T>::type;
+            using MedianResT = typename std::conditional<is_float, T, std::nullptr_t>::type;
+            using WM2D_IMPL = WaveletMatrix2dCu5C<WM_T, CH_NUM, WaveletMatrixMultiCu4G<XIdxT, 512>, 512, WORD_SIZE>;
+
+            CV_Assert(src.cols == dst.cols);
+            CV_Assert(dst.step % sizeof(T) == 0);
+
+            WM2D_IMPL WM_cuda(src.rows, src.cols, is_float, false);
+            WM_cuda.res_cu =  reinterpret_cast<WM_T*>(dst.ptr());
+
+            const size_t line_num = src.cols * CH_NUM;
+            if (is_float) {
+                WMMedianFloatSupporter::WMMedianFloatSupporter<float, CH_NUM, XYIdxT> float_supporter(src.rows, src.cols);
+                float_supporter.alloc();
+                for (int y = 0; y < src.rows; ++y) {
+                    cudaMemcpy(float_supporter.val_in_cu + y * line_num, src.ptr(y), line_num * sizeof(T), cudaMemcpyDeviceToDevice);
+                }
+                const auto p = WM_cuda.get_nowcu_and_buf_byte_div32();
+                float_supporter.sort_and_set((XYIdxT*)p.first, p.second);
+                WM_cuda.construct(nullptr, stream, true);
+                WM_cuda.template median2d<ThW, ThH, MedianResT, false>(radius, dst.step / sizeof(T), (MedianResT*)float_supporter.get_res_table(), stream);
+            } else {
+                for (int y = 0; y < src.rows; ++y) {
+                    cudaMemcpy(WM_cuda.src_cu + y * line_num, src.ptr(y), line_num * sizeof(T), cudaMemcpyDeviceToDevice);
+                }
+                WM_cuda.construct(nullptr, stream);
+                WM_cuda.template median2d<ThW, ThH, MedianResT, false>(radius, dst.step / sizeof(T), nullptr, stream);
+            }
+            WM_cuda.res_cu = nullptr;
+            if (!stream) {
+                cudaSafeCall( cudaDeviceSynchronize() );
+            }
+        }
+
+        template<typename T>
+        void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<T> src, PtrStepSz<T> dst, int radius, const int num_channels, cudaStream_t stream){
+            if (num_channels == 1) {
+                medianFiltering_wavelet_matrix_gpu<1>(src, dst, radius, stream);
+            } else if (num_channels == 3) {
+                medianFiltering_wavelet_matrix_gpu<3>(src, dst, radius, stream);
+            } else if (num_channels == 4) {
+                medianFiltering_wavelet_matrix_gpu<4>(src, dst, radius, stream);
+            } else {
+                CV_Assert(num_channels == 1 || num_channels == 3 || num_channels == 4);
+            }
+        }
+
+        template void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<uint8_t>  src, PtrStepSz<uint8_t>  dst, int radius, const int num_channels, cudaStream_t stream);
+        template void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<uint16_t> src, PtrStepSz<uint16_t> dst, int radius, const int num_channels, cudaStream_t stream);
+        template void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<float>    src, PtrStepSz<float>    dst, int radius, const int num_channels, cudaStream_t stream);
+}}}
+#endif // __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+
 #endif
diff --git a/modules/cudafilters/src/cuda/wavelet_matrix_2d.cuh b/modules/cudafilters/src/cuda/wavelet_matrix_2d.cuh
new file mode 100644
index 00000000000..9c10c223d87
--- /dev/null
+++ b/modules/cudafilters/src/cuda/wavelet_matrix_2d.cuh
@@ -0,0 +1,1053 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_WAVELET_MATRIX_2D_CUH__
+#define __OPENCV_WAVELET_MATRIX_2D_CUH__
+
+
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#include "wavelet_matrix_feature_support_checks.h"
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+
+
+#include <cub/cub.cuh>
+#include <vector>
+#include "opencv2/core/cuda/warp_shuffle.hpp"
+
+#include "wavelet_matrix_multi.cuh"
+
+namespace cv { namespace cuda { namespace device
+{
+
+namespace wavelet_matrix_median {
+    using std::vector;
+    using namespace std;
+
+template<int CH_NUM, int ThreadsDimY, int SRC_CACHE_DIV, int SRCB_S, typename SrcT, typename DstT, typename BlockT, typename XYIdxT, typename XIdxT>
+__global__ void WaveletMatrix2dCu5C_UpSweep_gpu(const SrcT mask, const uint16_t block_pair_num, const XYIdxT size_div_w, const SrcT* __restrict__ src, DstT* __restrict__ dst, BlockT* __restrict__ nbit_bp, const XYIdxT* __restrict__ nsum_buf_test, XYIdxT* __restrict__ nsum_buf_test2, const uint32_t bv_block_byte_div32, const uint32_t buf_byte_div32, const XIdxT* __restrict__ idx_p, const XIdxT inf, XIdxT* __restrict__ wm, XIdxT* __restrict__ nxt_idx, XYIdxT* __restrict__ wm_nsum_scan_buf, const XYIdxT cwm_buf_byte_div32, BlockT* __restrict__ nbit_bp_pre) {
+    using WordT = decltype(BlockT::nbit);
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+    constexpr int WORD_SIZE = 8 * sizeof(WordT);
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "");
+    constexpr uint32_t WORD_DIV_WARP = WORD_SIZE / WARP_SIZE;
+
+    static_assert(ThreadsDimY % SRC_CACHE_DIV == 0, "");
+    static_assert(ThreadsDimY != SRC_CACHE_DIV, "Warning: It's not efficient.");
+
+    const size_t buf_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * (buf_byte_div32*32ull);
+    const size_t bv_block_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * (bv_block_byte_div32*32ull);
+    const size_t cwm_buf_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * cwm_buf_byte_div32 * 32u;
+    src = (SrcT*)((uint8_t*)src + buf_byte_y_offset);
+    dst = (DstT*)((uint8_t*)dst + buf_byte_y_offset);
+    nsum_buf_test = (XYIdxT*)((uint8_t*)nsum_buf_test + buf_byte_y_offset);
+    nsum_buf_test2 = (XYIdxT*)((uint8_t*)nsum_buf_test2 + buf_byte_y_offset);
+    nbit_bp = (BlockT*)((uint8_t*)nbit_bp + bv_block_byte_y_offset);
+    nbit_bp_pre = (BlockT*)((uint8_t*)nbit_bp_pre + bv_block_byte_y_offset);
+
+    idx_p   = (XIdxT*)((uint8_t*)idx_p + buf_byte_y_offset);
+    nxt_idx = (XIdxT*)((uint8_t*)nxt_idx + buf_byte_y_offset);
+    if (wm != nullptr) wm = (XIdxT*)((uint8_t*)wm + cwm_buf_byte_y_offset);
+    wm_nsum_scan_buf = (XYIdxT*)((uint8_t*)wm_nsum_scan_buf + cwm_buf_byte_y_offset);
+
+
+    using WarpScanX = cub::WarpScan<XYIdxT, WARP_SIZE / SRC_CACHE_DIV>;
+    using WarpScanY = cub::WarpScan<XYIdxT, ThreadsDimY>;
+    using WarpReduce = cub::WarpReduce<uint32_t>;
+    using WarpReduceY = cub::WarpReduce<uint32_t, ThreadsDimY>;
+
+    static_assert(SRCB_S < 64 * 1024, "");
+
+    __shared__ SrcT   src_val_cache[ThreadsDimY][(WARP_SIZE/SRC_CACHE_DIV)-1][WARP_SIZE];
+    __shared__ XIdxT vidx_val_cache[ThreadsDimY][(WARP_SIZE/SRC_CACHE_DIV)-1][WARP_SIZE];
+
+    __shared__ uint4 nsum_count_sh[ThreadsDimY];
+    __shared__ XYIdxT wm_zero_count_sh[ThreadsDimY];
+    __shared__ XYIdxT pre_sum_share[2];
+    __shared__ XYIdxT warp_scan_sums[ThreadsDimY];
+    __shared__ typename WarpScanX::TempStorage s_scanStorage;
+    __shared__ typename WarpScanY::TempStorage s_scanStorage2;
+    __shared__ typename WarpReduce::TempStorage WarpReduce_temp_storage[ThreadsDimY];
+    __shared__ typename WarpReduceY::TempStorage WarpReduceY_temp_storage;
+    // shmem ------ end ------
+
+    XYIdxT wm_zero_count = 0;
+
+    const XYIdxT size_div_warp = size_div_w * WORD_DIV_WARP;
+    const XYIdxT nsum = nbit_bp[size_div_w].nsum;
+    const XYIdxT nsum_offset = nsum_buf_test[blockIdx.x];
+    const XYIdxT nsum_pre = nbit_bp_pre[size_div_w].nsum;
+
+
+    XYIdxT nsum_idx0_org = nsum_offset;
+    XYIdxT nsum_idx1_org = (XYIdxT)blockIdx.x * block_pair_num * THREAD_PER_GRID + nsum - nsum_idx0_org;
+    nsum_idx0_org /= (XYIdxT)block_pair_num * ThreadsDimY * WARP_SIZE;
+    nsum_idx1_org /= (XYIdxT)block_pair_num * ThreadsDimY * WARP_SIZE;
+    const XYIdxT nsum_idx0_bound = (nsum_idx0_org + 1) * block_pair_num * ThreadsDimY * WARP_SIZE;
+    const XYIdxT nsum_idx1_bound = (nsum_idx1_org + 1) * block_pair_num * ThreadsDimY * WARP_SIZE;
+    uint4 nsum_count = make_uint4(0, 0, 0, 0);
+
+    const unsigned short th_idx = threadIdx.y * WARP_SIZE + threadIdx.x;
+    if (th_idx == 0) {
+        pre_sum_share[0] = nsum_offset;
+    }
+
+    for (XYIdxT ka = 0; ka < block_pair_num; ka += WARP_SIZE / SRC_CACHE_DIV) {
+        const XYIdxT ibb = ((XYIdxT)blockIdx.x * block_pair_num + ka) * ThreadsDimY;
+        if (ibb >= size_div_warp) break;
+
+        WarpWT my_bits = 0;
+        SrcT  first_val;
+        XIdxT first_idxval;
+
+        for (XYIdxT kb = 0, i = ibb + WARP_SIZE / SRC_CACHE_DIV * threadIdx.y; kb < WARP_SIZE / SRC_CACHE_DIV; ++kb, ++i) {
+            if (i >= size_div_warp) break;
+            WarpWT bits;
+            const XYIdxT ij = i * WARP_SIZE + threadIdx.x;
+            const SrcT v = src[ij];
+            const XIdxT wm_idxv = idx_p[ij];
+            if (kb == 0) {
+                first_val = v;
+                first_idxval = wm_idxv;
+            } else {
+                src_val_cache[threadIdx.y][kb - 1][threadIdx.x] = v;
+                vidx_val_cache[threadIdx.y][kb - 1][threadIdx.x] = wm_idxv;
+            }
+            if (v <= mask) {
+                bits = __activemask();
+            } else {
+                bits = ~__activemask();
+            }
+            if (threadIdx.x == kb) {
+                my_bits = bits;
+            }
+            if (wm != nullptr) {
+                if (ij < nsum_pre) {
+                    wm[ij] = wm_idxv;
+                    if (wm_idxv * 2 <= inf) {
+                        ++wm_zero_count;
+                    }
+                } else {
+                    wm[ij] = inf;
+                }
+            }
+        }
+
+        XYIdxT c, t = 0;
+        if (threadIdx.y < ThreadsDimY) {
+            c = __popc(my_bits);
+
+            WarpScanX(s_scanStorage).ExclusiveSum(c, t);
+            if (threadIdx.x == WARP_SIZE / SRC_CACHE_DIV - 1) {
+                warp_scan_sums[threadIdx.y] = c + t;
+            }
+        }
+
+        __syncthreads();
+
+        XYIdxT pre_sum = pre_sum_share[(ka & (WARP_SIZE / SRC_CACHE_DIV)) > 0 ? 1 : 0];
+        XYIdxT s = threadIdx.x < ThreadsDimY ? warp_scan_sums[threadIdx.x] : 0;
+        WarpScanY(s_scanStorage2).ExclusiveSum(s, s);
+
+        s = cv::cuda::device::shfl(s, threadIdx.y, WARP_SIZE);
+        s += t + pre_sum;
+
+        if (SRC_CACHE_DIV == 1 || threadIdx.x < WARP_SIZE / SRC_CACHE_DIV) {
+            if (th_idx == THREAD_PER_GRID - WARP_SIZE + WARP_SIZE / SRC_CACHE_DIV - 1) {
+                pre_sum_share[(ka & (WARP_SIZE / SRC_CACHE_DIV)) == 0 ? 1 : 0] = s + c;
+            }
+            const XYIdxT bi = ibb + threadIdx.y * WARP_SIZE / SRC_CACHE_DIV + threadIdx.x;
+            if (bi < size_div_warp) {
+                static_assert(WORD_SIZE == 32, "");
+                nbit_bp[bi] = BlockT{s, my_bits};
+            }
+        }
+
+        if (mask == 0) {
+            SrcT  vo = first_val;
+            XIdxT idx_v = first_idxval;
+            for (XYIdxT j = 0, i = ibb + WARP_SIZE / SRC_CACHE_DIV * threadIdx.y; j < WARP_SIZE / SRC_CACHE_DIV; ++j, ++i) {
+                if (i >= size_div_warp) break;
+                const WarpWT e_nbit = cv::cuda::device::shfl(my_bits, j, WARP_SIZE);
+                const XYIdxT e_nsum = cv::cuda::device::shfl(s, j, WARP_SIZE);
+                XYIdxT rank = __popc(e_nbit << (WARP_SIZE - threadIdx.x));
+                const XYIdxT idx0 = e_nsum + rank;
+                XYIdxT idx = idx0;
+                if (vo > mask) { // 1
+                    const XYIdxT ij = i * WARP_SIZE + threadIdx.x;
+                    idx = ij + nsum - idx;
+                }
+                if (idx < size_div_warp * WARP_SIZE) {
+                    nxt_idx[idx] = idx_v;
+                }
+                if (j == WARP_SIZE / SRC_CACHE_DIV - 1) break;
+                vo    = src_val_cache[threadIdx.y][j][threadIdx.x];
+                idx_v = vidx_val_cache[threadIdx.y][j][threadIdx.x];
+            }
+            continue;
+        }
+        const SrcT mask_2 = mask >> 1;
+        SrcT  vo = first_val;
+        XIdxT idx_v = first_idxval;
+        for (XYIdxT j = 0, i = ibb + WARP_SIZE / SRC_CACHE_DIV * threadIdx.y; j < WARP_SIZE / SRC_CACHE_DIV; ++j, ++i) {
+            if (i >= size_div_warp) break;
+            const WarpWT e_nbit = cv::cuda::device::shfl(my_bits, j, WARP_SIZE);
+            const XYIdxT e_nsum = cv::cuda::device::shfl(s, j, WARP_SIZE);
+            XYIdxT rank = __popc(e_nbit << (WARP_SIZE - threadIdx.x));
+            const XYIdxT idx0 = e_nsum + rank;
+
+            DstT v = (DstT)vo;
+            XYIdxT idx = idx0;
+            if (vo > mask) { // 1
+                const XYIdxT ij = i * WARP_SIZE + threadIdx.x;
+                idx = ij + nsum - idx;
+                v &= mask;
+            }
+            if (idx < size_div_warp * WARP_SIZE) {
+                if (mask != 0) {
+                    dst[idx] = v;
+                }
+                nxt_idx[idx] = idx_v;
+            }
+
+            if (v <= mask_2) {
+                if (vo <= mask) {
+                    if (idx < nsum_idx0_bound) {
+                        nsum_count.x++;
+                    } else {
+                        nsum_count.y++;
+                    }
+                } else {
+                    if (idx < nsum_idx1_bound) {
+                        nsum_count.z++;
+                    } else {
+                        nsum_count.w++;
+                    }
+                }
+            }
+            if (j == WARP_SIZE / SRC_CACHE_DIV - 1) break;
+            vo    = src_val_cache[threadIdx.y][j][threadIdx.x];
+            idx_v = vidx_val_cache[threadIdx.y][j][threadIdx.x];
+        }
+    }
+    if (blockIdx.x == gridDim.x - 1 && th_idx == 0) {
+        nbit_bp[size_div_warp / WORD_DIV_WARP].nsum = nsum;
+    }
+
+    nsum_count.x = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.x);
+    nsum_count.y = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.y);
+    nsum_count.z = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.z);
+    nsum_count.w = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.w);
+    wm_zero_count = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(wm_zero_count);
+    if (threadIdx.x == 0) {
+        nsum_count_sh[threadIdx.y]    = nsum_count;
+        wm_zero_count_sh[threadIdx.y] = wm_zero_count;
+    }
+    __syncthreads();
+    if (threadIdx.x < ThreadsDimY) {
+        nsum_count = nsum_count_sh[threadIdx.x];
+        nsum_count.x = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.x);
+        nsum_count.y = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.y);
+        nsum_count.z = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.z);
+        nsum_count.w = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.w);
+        wm_zero_count = WarpReduceY(WarpReduceY_temp_storage).Sum(wm_zero_count_sh[threadIdx.x]);
+        if (th_idx == 0) {
+            const XYIdxT nsum_idx0_org = nsum_idx0_bound / ((XYIdxT)block_pair_num * ThreadsDimY * WARP_SIZE);
+            const XYIdxT nsum_idx1_org = nsum_idx1_bound / ((XYIdxT)block_pair_num * ThreadsDimY * WARP_SIZE);
+            if (mask != 0) {
+                if (nsum_count.x > 0) atomicAdd(nsum_buf_test2 + nsum_idx0_org - 1, nsum_count.x);
+                if (nsum_count.y > 0) atomicAdd(nsum_buf_test2 + nsum_idx0_org - 0, nsum_count.y);
+                if (nsum_count.z > 0) atomicAdd(nsum_buf_test2 + nsum_idx1_org - 1, nsum_count.z);
+                if (nsum_count.w > 0) atomicAdd(nsum_buf_test2 + nsum_idx1_org - 0, nsum_count.w);
+            }
+            if (wm != nullptr) {
+                wm_nsum_scan_buf[blockIdx.x] = wm_zero_count;
+            }
+        }
+    }
+}
+
+template<int CH_NUM, int ThreadsDimY, typename BlockT, typename XYIdxT, typename XIdxT>
+__global__ void WaveletMatrix2dCu5C_last_gpu(const uint16_t block_pair_num, const XYIdxT size_div_w, const uint32_t buf_byte_div32, const XIdxT* __restrict__ idx_p, const XIdxT inf, XIdxT* __restrict__ wm, XYIdxT* __restrict__ wm_nsum_scan_buf, const XYIdxT cwm_buf_byte_div32, BlockT* __restrict__ nbit_bp_pre, const uint32_t bv_block_byte_div32) {
+    using WordT = decltype(BlockT::nbit);
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+    constexpr int WORD_SIZE = 8 * sizeof(WordT);
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "");
+    constexpr uint32_t WORD_DIV_WARP = WORD_SIZE / WARP_SIZE;
+
+    const size_t buf_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * (buf_byte_div32*32ull);
+    const size_t bv_block_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * (bv_block_byte_div32*32ull);
+    const size_t cwm_buf_byte_y_offset = (size_t)(CH_NUM==1?0:blockIdx.y) * cwm_buf_byte_div32 * 32u;
+
+    idx_p   = (XIdxT*)((uint8_t*)idx_p + buf_byte_y_offset);
+    wm = (XIdxT*)((uint8_t*)wm + cwm_buf_byte_y_offset);
+    wm_nsum_scan_buf = (XYIdxT*)((uint8_t*)wm_nsum_scan_buf + cwm_buf_byte_y_offset);
+    nbit_bp_pre = (BlockT*)((uint8_t*)nbit_bp_pre + bv_block_byte_y_offset);
+    const XYIdxT nsum_pre = nbit_bp_pre[size_div_w].nsum;
+
+    using WarpReduce = cub::WarpReduce<uint32_t>;
+    using WarpReduceY = cub::WarpReduce<uint32_t, ThreadsDimY>;
+
+    __shared__ XYIdxT wm_zero_count_sh[ThreadsDimY];
+    __shared__ typename WarpReduce::TempStorage WarpReduce_temp_storage[ThreadsDimY];
+    __shared__ typename WarpReduceY::TempStorage WarpReduceY_temp_storage;
+    // shmem ------ end ------
+
+    XYIdxT wm_zero_count = 0;
+    const XYIdxT size_div_warp = size_div_w * WORD_DIV_WARP;
+    const unsigned short th_idx = threadIdx.y * WARP_SIZE + threadIdx.x;
+
+    const int block_num = block_pair_num/WARP_SIZE;
+    for (XYIdxT ka = 0; ka < block_num; ++ka) {
+        const XYIdxT ibb = ((XYIdxT)blockIdx.x * block_num + ka) * THREAD_PER_GRID + WARP_SIZE * threadIdx.y;
+        if (ibb >= size_div_warp) break;
+
+        for (XYIdxT kb = 0; kb < WARP_SIZE; ++kb) {
+            XYIdxT i = ibb + kb;
+            if (i >= size_div_warp) break;
+
+            const XYIdxT ij = i * WARP_SIZE + threadIdx.x;
+
+            if (ij < nsum_pre) {
+                const XIdxT wm_idxv = idx_p[ij];
+                wm[ij] = wm_idxv;
+                if (wm_idxv * 2 <= inf) {
+                    ++wm_zero_count;
+                }
+            } else {
+                wm[ij] = inf;
+            }
+        }
+    }
+    wm_zero_count = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(wm_zero_count);
+    if (threadIdx.x == 0) {
+        wm_zero_count_sh[threadIdx.y] = wm_zero_count;
+    }
+    __syncthreads();
+    if (threadIdx.x < ThreadsDimY) {
+        wm_zero_count = WarpReduceY(WarpReduceY_temp_storage).Sum(wm_zero_count_sh[threadIdx.x]);
+        if (th_idx == 0) {
+            wm_nsum_scan_buf[blockIdx.x] = wm_zero_count;
+        }
+    }
+}
+
+template<int CH_NUM, int MAX_BLOCK_X, typename XYIdxT, typename BlockT>
+__global__ void WaveletMatrix2dCu5C_ExclusiveSum(XYIdxT* __restrict__ nsum_scan_buf, XYIdxT* __restrict__ nsum_buf_test2, BlockT* __restrict__ nsum_p, const uint32_t buf_byte_div32, const uint32_t bv_block_byte_div32) {
+
+    typedef cub::BlockScan<XYIdxT, MAX_BLOCK_X> BlockScan;
+    __shared__ typename BlockScan::TempStorage temp_storage;
+
+    XYIdxT thread_data1;
+    XYIdxT thread_data2;
+
+    nsum_scan_buf = (XYIdxT*)((uint8_t*)nsum_scan_buf + (size_t)blockIdx.x * (buf_byte_div32*32ull));
+    nsum_buf_test2 = (XYIdxT*)((uint8_t*)nsum_buf_test2 + (size_t)blockIdx.x * (buf_byte_div32*32ull));
+
+    thread_data1 = nsum_scan_buf[threadIdx.x];
+    BlockScan(temp_storage).ExclusiveSum(thread_data1, thread_data2);
+
+    nsum_scan_buf[threadIdx.x] = thread_data2;
+    nsum_buf_test2[threadIdx.x] = 0;
+
+    if (threadIdx.x == blockDim.x - 1) {
+        thread_data2 += thread_data1;
+        nsum_p = (BlockT*)((uint8_t*)nsum_p + (size_t)blockIdx.x * (bv_block_byte_div32*32ull));
+        nsum_p->nsum = thread_data2;
+    }
+}
+
+
+template<int CH_NUM, int ThreadsDimY, bool MOV_SRC,typename SrcT, typename XYIdxT, typename XIdxT>
+__global__ void WaveletMatrix2dCu5C_first_gpu_multi(const SrcT mask, uint16_t block_pair_num, const XYIdxT size_div_warp, SrcT* __restrict__ src, SrcT* __restrict__ dst, XYIdxT* __restrict__ nsum_scan_buf, const uint32_t buf_byte_div32, XIdxT* __restrict__ buf_idx, const int W, const XYIdxT WH, const SrcT* __restrict__ src_const) {
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+
+    SrcT* __restrict__ dsts[CH_NUM];
+    XIdxT* __restrict__ buf_idxs[CH_NUM];
+    XYIdxT cs[CH_NUM];
+
+    __shared__ SrcT src_vbuf_org[ThreadsDimY][CH_NUM * WARP_SIZE];
+    SrcT* __restrict__ src_vbuf = src_vbuf_org[threadIdx.y];
+
+    for (int c = 0; c < CH_NUM; ++c) {
+        if (CH_NUM > 1) { // constexpr
+            dsts[c]     = (SrcT*)((uint8_t*)dst + (size_t)c * (buf_byte_div32*32ull));
+        }
+        buf_idxs[c] = (XIdxT*)((uint8_t*)buf_idx + (size_t)c * (buf_byte_div32*32ull));
+        cs[c] = 0;
+    }
+
+    XYIdxT ibb = (XYIdxT)blockIdx.x * block_pair_num * ThreadsDimY;
+    for (XYIdxT ka = 0; ka < block_pair_num; ka += WARP_SIZE, ibb += THREAD_PER_GRID) {
+        for (XYIdxT kb = 0, i = ibb + WARP_SIZE * threadIdx.y; kb < WARP_SIZE; ++kb, ++i) {
+            if (i >= size_div_warp) break;
+            const XYIdxT iw = i * WARP_SIZE;
+            const XYIdxT idx = iw + threadIdx.x;
+            const XIdxT idx_v = (idx >= WH ? 0 : idx % W);
+
+            for (int c = 0; c < CH_NUM; ++c) {
+                if (MOV_SRC) { // constexpr
+                    const XYIdxT s_idx = iw * CH_NUM + threadIdx.x + c * WARP_SIZE;
+                    src_vbuf[threadIdx.x + c * WARP_SIZE] = src[s_idx] = src_const[s_idx];
+                } else {
+                    src_vbuf[threadIdx.x + c * WARP_SIZE] = src[iw * CH_NUM + threadIdx.x + c * WARP_SIZE];
+                }
+            }
+            __syncwarp();
+            for (int c = 0; c < CH_NUM; ++c) {
+                const SrcT v = src_vbuf[threadIdx.x * CH_NUM + c];
+                if (v <= mask) {
+                    ++cs[c];
+                }
+                buf_idxs[c][idx] = idx_v;
+                if constexpr(CH_NUM > 1) {
+                    dsts[c][idx] = v;
+                }
+            }
+        }
+    }
+
+    using WarpReduce = cub::WarpReduce<uint32_t>;
+    __shared__ typename WarpReduce::TempStorage WarpReduce_temp_storage[ThreadsDimY];
+    __shared__ XYIdxT cs_sum_sh[CH_NUM][ThreadsDimY];
+    for (int c = 0; c < CH_NUM; ++c) {
+        cs[c] = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(cs[c]);
+    }
+    if (threadIdx.x == 0) {
+        for (int c = 0; c < CH_NUM; ++c) {
+            cs_sum_sh[c][threadIdx.y] = cs[c];
+        }
+    }
+    __syncthreads();
+    if (threadIdx.y != 0) return;
+    for (int c = 0; c < CH_NUM; ++c) {
+        XYIdxT cs_bsum = (threadIdx.x < ThreadsDimY ? cs_sum_sh[c][threadIdx.x] : 0);
+        cs_bsum = WarpReduce(WarpReduce_temp_storage[0]).Sum(cs_bsum);
+        if (threadIdx.x == 0) {
+            nsum_scan_buf[blockIdx.x] = cs_bsum;
+            nsum_scan_buf = (XYIdxT*)((uint8_t*)nsum_scan_buf + (buf_byte_div32*32ull));
+        }
+    }
+
+}
+
+
+template<int ThreadsDimY, typename SrcT, typename XYIdxT, typename XIdxT>
+__global__ void WaveletMatrix2dCu5C_first_gpu_multi_srcunpacked(const SrcT mask, uint16_t block_pair_num, const XYIdxT size_div_warp, const SrcT* __restrict__ src, XYIdxT* __restrict__ nsum_scan_buf, const uint32_t buf_byte_div32, XIdxT* __restrict__ buf_idx, const int W, const XYIdxT WH) {
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+
+    XYIdxT cs = 0;
+
+    const int c = blockIdx.y;
+    buf_idx = buf_idx + c * (buf_byte_div32*32ull / sizeof(XIdxT));
+    src = src + c * (buf_byte_div32*32ull / sizeof(SrcT));
+
+
+    XYIdxT i = (XYIdxT)blockIdx.x * block_pair_num * ThreadsDimY + threadIdx.y;
+
+    XIdxT x_idx = (i * WARP_SIZE + threadIdx.x) % W;
+    const XIdxT x_diff = THREAD_PER_GRID % W;
+
+    for (XYIdxT k = 0; k < block_pair_num; ++k, i += ThreadsDimY) {
+        if (i >= size_div_warp) break;
+        const XYIdxT idx = i * WARP_SIZE + threadIdx.x;
+
+        if (idx >= WH) x_idx = 0;
+
+        const SrcT v = src[idx];
+        if (v <= mask) {
+            ++cs;
+        }
+        buf_idx[idx] = x_idx;
+
+        x_idx += x_diff;
+        if (x_idx >= W) x_idx -= W;
+    }
+
+    using WarpReduce = cub::WarpReduce<uint32_t>;
+    __shared__ typename WarpReduce::TempStorage WarpReduce_temp_storage[ThreadsDimY];
+    __shared__ XYIdxT cs_sum_sh[ThreadsDimY];
+    cs = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(cs);
+    if (threadIdx.x == 0) {
+        cs_sum_sh[threadIdx.y] = cs;
+    }
+    __syncthreads();
+    if (threadIdx.y != 0) return;
+    XYIdxT cs_bsum = (threadIdx.x < ThreadsDimY ? cs_sum_sh[threadIdx.x] : 0);
+    cs_bsum = WarpReduce(WarpReduce_temp_storage[0]).Sum(cs_bsum);
+    if (threadIdx.x == 0) {
+        nsum_scan_buf += (buf_byte_div32*32ull / sizeof(XYIdxT)) * (blockIdx.y);
+        nsum_scan_buf[blockIdx.x] = cs_bsum;
+    }
+}
+
+template<typename IdxType, typename BlockT>
+__device__
+inline IdxType WaveletMatrix2dCu5C_median2d_rank0(const IdxType i, const BlockT* __restrict__ nbit_bp) {
+    using WordT = decltype(BlockT::nbit);
+    constexpr int WORD_SIZE = 8 * sizeof(WordT);
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "");
+
+    const IdxType bi = i / WORD_SIZE;
+
+    const int ai = i % WORD_SIZE;
+    const BlockT block = nbit_bp[bi];
+    if constexpr(WORD_SIZE == 32) {
+        return block.nsum + __popc(block.nbit & ((1u << ai) - 1));
+    }
+    if constexpr(WORD_SIZE == 64) {
+        return block.nsum + __popcll(block.nbit & ((1ull << ai) - 1ull));
+    }
+}
+
+
+template<int CH_NUM, int THREADS_NUM_W, int THREADS_NUM_H, int VAL_BIT_LEN, bool CUT_BORDER, typename XYIdxT, typename XIdxT, typename ValT, typename BlockT, typename ResT, typename ResTableT>
+__global__ void WaveletMatrix2dCu5C_median2d_cu(
+    const int H, const int W, const int res_step_num, const int r, ResT* __restrict__ res_cu, const BlockT* __restrict__ wm_nbit_bp, const uint32_t nsum_pos,
+    const uint32_t bv_block_h_byte_div32, const uint32_t bv_block_len,
+    const BlockT* __restrict__ bv_nbit_bp, const uint8_t w_bit_len, const uint8_t val_bit_len,
+    const ResTableT* __restrict__ res_table
+) {
+
+    const int y = blockIdx.y * THREADS_NUM_H + threadIdx.y;
+    if (y >= H) return;
+    const int x = blockIdx.x * THREADS_NUM_W + threadIdx.x;
+    if (x >= W) return;
+
+    if (CH_NUM >= 2) { // constexpr
+        bv_nbit_bp = (BlockT*)((uint8_t*)bv_nbit_bp + bv_block_h_byte_div32 * 32ull * blockIdx.z * (VAL_BIT_LEN >= 0 ? VAL_BIT_LEN : val_bit_len)); // TODO
+        wm_nbit_bp = (BlockT*)((uint8_t*)wm_nbit_bp + bv_block_h_byte_div32 * 32ull * blockIdx.z * w_bit_len);
+    }
+
+    XYIdxT ya, yb, k;
+    XIdxT  xa, xb;
+    if (CUT_BORDER) { // constexpr
+        ya = y;
+        xa = x;
+        yb = y + r * 2 + 1;
+        xb = x + r * 2 + 1;
+        k = (r * 2 + 1) * (r * 2 + 1) / 2;
+    } else {
+        ya = (y < r ? 0 : y - r);
+        xa = (x < r ? 0 : x - r);
+        yb = y + r + 1; if (yb > H) yb = H;
+        xb = x + r + 1; if (xb > W) xb = W;
+        k = XYIdxT(yb - ya) * (xb - xa) / 2;
+    }
+    ValT res = 0;
+    ya *= (CUT_BORDER ? W + 2 * r : W);
+    yb *= (CUT_BORDER ? W + 2 * r : W);
+
+    for (int8_t h = (VAL_BIT_LEN >= 0 ? VAL_BIT_LEN : val_bit_len); h--; ) {
+        const XYIdxT top0 = WaveletMatrix2dCu5C_median2d_rank0(ya, bv_nbit_bp);
+        const XYIdxT bot0 = WaveletMatrix2dCu5C_median2d_rank0(yb, bv_nbit_bp);
+        XYIdxT l_ya_xa = top0;
+        XYIdxT l_yb_xa = bot0;
+        XYIdxT l_ya_xb = top0;
+        XYIdxT l_yb_xb = bot0;
+        XYIdxT d = 0;
+        for (int8_t j = w_bit_len; j--; ) {
+            const XYIdxT zeros = wm_nbit_bp[nsum_pos].nsum;
+            const XYIdxT l_ya_xa_rank0 = WaveletMatrix2dCu5C_median2d_rank0(l_ya_xa, wm_nbit_bp);
+            const XYIdxT l_ya_xb_rank0 = WaveletMatrix2dCu5C_median2d_rank0(l_ya_xb, wm_nbit_bp);
+            const XYIdxT l_yb_xb_rank0 = WaveletMatrix2dCu5C_median2d_rank0(l_yb_xb, wm_nbit_bp);
+            const XYIdxT l_yb_xa_rank0 = WaveletMatrix2dCu5C_median2d_rank0(l_yb_xa, wm_nbit_bp);
+
+            if (((xa >> j) & 1) == 0) {
+                l_ya_xa = l_ya_xa_rank0;
+                l_yb_xa = l_yb_xa_rank0;
+            } else {
+                d += l_ya_xa_rank0; l_ya_xa += zeros - l_ya_xa_rank0;
+                d -= l_yb_xa_rank0; l_yb_xa += zeros - l_yb_xa_rank0;
+            }
+            if (((xb >> j) & 1) == 0) {
+                l_ya_xb = l_ya_xb_rank0;
+                l_yb_xb = l_yb_xb_rank0;
+            } else {
+                d -= l_ya_xb_rank0; l_ya_xb += zeros - l_ya_xb_rank0;
+                d += l_yb_xb_rank0; l_yb_xb += zeros - l_yb_xb_rank0;
+            }
+            wm_nbit_bp = (BlockT*)((uint8_t*)wm_nbit_bp - bv_block_h_byte_div32 * 32ull);
+        }
+        if (CH_NUM >= 2) {
+            wm_nbit_bp = (BlockT*)((uint8_t*)wm_nbit_bp - bv_block_h_byte_div32 * 32ull * w_bit_len * (CH_NUM - 1));
+        }
+        const XYIdxT bv_h_zeros = bv_nbit_bp[nsum_pos].nsum;
+        if (k < d) {
+            ya = top0;
+            yb = bot0;
+        } else {
+            k -= d;
+            res |= (ValT)1 << h;
+            ya += bv_h_zeros - top0;
+            yb += bv_h_zeros - bot0;
+        }
+        bv_nbit_bp = (BlockT*)((uint8_t*)bv_nbit_bp - bv_block_h_byte_div32 * 32ull);
+    }
+
+
+
+
+    if constexpr(is_same<ResTableT, std::nullptr_t>::value) {
+        res_cu[(XYIdxT)y * res_step_num + x * CH_NUM + blockIdx.z] = res;
+    } else if (CH_NUM == 1){
+        res_cu[(XYIdxT)y * res_step_num + x * CH_NUM] = res_table[res];
+    } else {
+        const size_t offset = size_t(CUT_BORDER ? W + 2 * r : W) * (CUT_BORDER ? H + 2 * r : H) * blockIdx.z;
+        res_cu[(XYIdxT)y * res_step_num + x * CH_NUM + blockIdx.z] = res_table[res + offset];
+    }
+}
+
+
+
+
+template <typename ValT, int CH_NUM, class MultiWaveletMatrixImpl, int TH_NUM = 512, int WORD_SIZE = 32>
+struct WaveletMatrix2dCu5C {
+    static_assert(is_same<ValT, uint32_t>() || is_same<ValT, uint16_t>() || is_same<ValT, uint8_t>(), "Supports 32, 16, or 8 bits only");
+    static constexpr int MAX_BIT_LEN = 8 * sizeof(ValT);
+
+    static constexpr uint32_t WSIZE = WORD_SIZE;
+    static constexpr int WARP_SIZE = 32;
+    using T_Type = ValT;
+    static constexpr int THREAD_PER_GRID = TH_NUM;
+    static constexpr int SRC_CACHE_DIV = 2;
+    static constexpr int MAX_BLOCK_X = MultiWaveletMatrixImpl::MAX_BLOCK_X;
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "WORD_SIZE must be 32 or 64");
+    using WordT = typename std::conditional<WORD_SIZE == 32, uint32_t, uint64_t>::type;
+
+    static_assert(MAX_BLOCK_X <= 1024, "");
+    static_assert(TH_NUM == 1024 || TH_NUM == 512 || TH_NUM == 256 || TH_NUM == 128 || TH_NUM == 64 || TH_NUM == 32, "");
+    static_assert(THREAD_PER_GRID == MultiWaveletMatrixImpl::THREAD_PER_GRID, "");
+
+    using BlockT = typename MultiWaveletMatrixImpl::BlockT;
+    using WarpWT = uint32_t;
+    using XIdxT = uint16_t;
+    using YIdxT = uint16_t;
+    using XYIdxT = uint32_t;
+    static constexpr int BLOCK_TYPE = 2;
+    using MultiWaveletMatrixImplClass = MultiWaveletMatrixImpl;
+    static_assert(is_same<XIdxT, typename MultiWaveletMatrixImpl::T_Type>::value, "");
+    static_assert(8 * sizeof(WarpWT) == WARP_SIZE, "");
+
+    int H, W;
+    XYIdxT size = 0;
+    MultiWaveletMatrixImpl WM;
+    XYIdxT bv_zeros[MAX_BIT_LEN];
+
+    int w_bit_len = 0;
+    int val_bit_len = 0;
+    static constexpr int wm_num = CH_NUM;
+
+private:
+    uint8_t* bv_block_nbit_and_nsum_base_cu = nullptr; // GPU mem
+    uint32_t bv_block_byte_div32;
+    uint32_t buf_byte_div32;
+    uint32_t nsum_scan_buf_len;
+    size_t input_buf_byte;
+public:
+    ValT* src_cu = nullptr; // GPU mem
+    ValT* res_cu = nullptr;
+    size_t bv_block_len = 0;
+    size_t bv_chunk_len = 0;
+
+#if _MSC_VER >= 1920 || __INTEL_COMPILER
+    inline static int bitCount64(uint64_t bits) {
+        return (int)_mm_popcnt_u64(bits);
+    }
+#else
+    inline static int bitCount64(uint64_t bits) {
+        return __builtin_popcountll(bits);
+    }
+#endif
+    static constexpr int get_bit_len(uint64_t val) {
+        return (
+            (val |= val >> 1),
+            (val |= val >> 2),
+            (val |= val >> 4),
+            (val |= val >> 8),
+            (val |= val >> 16),
+            (val |= val >> 32),
+            bitCount64(val));
+        // val |= val >> 1;
+        // val |= val >> 2;
+        // val |= val >> 4;
+        // val |= val >> 8;
+        // val |= val >> 16;
+        // val |= val >> 32;
+        // return bitCount64(val);
+    }
+
+    WaveletMatrix2dCu5C() {
+        reset(0, 0);
+    }
+    WaveletMatrix2dCu5C(const int rows, const int cols, const bool use_hw_bit_len = false, const bool alloc_res = true) {
+        reset(rows, cols, use_hw_bit_len, alloc_res);
+    }
+
+    void reset(const int rows, const int cols, const bool use_hw_bit_len = false, const bool alloc_res = true) {
+        H = rows;
+        W = cols;
+        if (rows == 0 || cols == 0) return;
+        val_bit_len = (use_hw_bit_len ? get_bit_len((uint64_t)H * W - 1) : MAX_BIT_LEN);
+        assert(size == 0 && src_cu == nullptr);
+
+        size = div_ceil<size_t>((uint64_t)H * W, WORD_SIZE) * WORD_SIZE;
+        assert(W < 65535); // That is, less than 65534.
+        w_bit_len = get_bit_len(W); // w=7 [0,6] bl=3; w=8 [0,7] bl=4
+        WM.reset(size, w_bit_len, val_bit_len * wm_num);
+        if (val_bit_len == 0) return;
+
+
+        bv_block_len = div_ceil<size_t>(size, THREAD_PER_GRID) * THREAD_PER_GRID / WORD_SIZE + 1;
+        bv_block_len = div_ceil<size_t>(bv_block_len, 8*2) * 8*2;
+        const size_t bv_block_byte = (sizeof(BlockT)) * val_bit_len * bv_block_len;
+        bv_block_byte_div32 = div_ceil<size_t>(bv_block_byte, 32);
+
+        cudaMalloc(&bv_block_nbit_and_nsum_base_cu, (size_t)(bv_block_byte_div32*32ull) * CH_NUM);
+        if (bv_block_nbit_and_nsum_base_cu == nullptr) { printf("GPU Memory Alloc Error! %s:%d\n", __FILE__, __LINE__); release(); return; }
+
+        const uint16_t block_pair_num = get_block_pair_num();
+        nsum_scan_buf_len = div_ceil<size_t>(size, (size_t)THREAD_PER_GRID * block_pair_num);
+        nsum_scan_buf_len = div_ceil<size_t>(nsum_scan_buf_len, 4) * 4;
+
+        const size_t buf_byte =
+            sizeof(XYIdxT) * 2 * nsum_scan_buf_len
+            + sizeof(XIdxT) * 2 * size
+            + sizeof(ValT) * (CH_NUM == 1 ? 1 : 2) * size;
+        buf_byte_div32 = div_ceil<size_t>(buf_byte, 32);
+
+
+        input_buf_byte = sizeof(ValT) * size * CH_NUM;
+        cudaMalloc(&src_cu, (size_t)(buf_byte_div32*32ull) * CH_NUM + input_buf_byte);
+        if (src_cu == nullptr) { printf("GPU Memory Alloc Error! %s:%d\n", __FILE__, __LINE__); release(); return; }
+
+        if (alloc_res) {
+            cudaMalloc(&res_cu, sizeof(ValT) * size * CH_NUM);
+            if (res_cu == nullptr) { printf("GPU Memory Alloc Error! %s:%d\n", __FILE__, __LINE__); release(); return; }
+        }
+    }
+    void release() {
+        size = 0;
+        if (src_cu != nullptr) cudaFree(src_cu);
+        if (bv_block_nbit_and_nsum_base_cu != nullptr) cudaFree(bv_block_nbit_and_nsum_base_cu);
+        if (res_cu != nullptr) cudaFree(res_cu);
+        src_cu = nullptr;
+        bv_block_nbit_and_nsum_base_cu = nullptr;
+        res_cu = nullptr;
+    }
+    ~WaveletMatrix2dCu5C() { release(); }
+
+    BlockT*  get_bv_block_cu(int h) const { return (BlockT*)(bv_block_nbit_and_nsum_base_cu + (bv_block_len * (sizeof(BlockT))) * h); }
+
+    BlockT*  get_bv_block_cu(int h, int c) const { return (BlockT*)((uint8_t*)get_bv_block_cu(h) + (size_t)c * (bv_block_byte_div32*32ull)); }
+
+
+    uint16_t get_block_pair_num() const {
+        return WM.get_block_pair_num() * MultiWaveletMatrixImpl::THREAD_PER_GRID / THREAD_PER_GRID;
+    }
+    std::pair<ValT*, XYIdxT> get_nowcu_and_buf_byte_div32() {
+        ValT *now_cu = src_cu + (CH_NUM == 1 ? 0ull : size * (size_t)CH_NUM);
+        return make_pair(now_cu, buf_byte_div32);
+    }
+
+    // Set data in src_cu before calling (data will be destroyed). Or set src_cu_const.
+    void construct(const ValT *src_cu_const = nullptr, const cudaStream_t main_stream = 0, const bool src_unpacked = false) {
+        assert(size > 0 && src_cu != nullptr);
+        if (val_bit_len == 0) return;
+        if (src_cu == nullptr) { printf("Build Error: memory not alloced."); return;}
+
+        const XIdxT inf = ((XIdxT)1u << w_bit_len) - 1;
+        assert(W <= inf);
+        assert(size % WORD_SIZE == 0);
+
+        ValT mask = ((ValT)1u << val_bit_len) - 1;
+
+        const uint16_t block_pair_num = get_block_pair_num();
+        const int grid_x = div_ceil<int>(size, THREAD_PER_GRID * block_pair_num);
+        if (grid_x > MAX_BLOCK_X) { printf("over grid_x %d\n", grid_x); exit(1); }
+
+        const dim3 grid(grid_x, wm_num);
+        const dim3 thread(WARP_SIZE, THREAD_PER_GRID / WARP_SIZE);
+        const XYIdxT size_div_w = size / WORD_SIZE;
+        const XYIdxT size_div_warp = size / WARP_SIZE;
+        assert(size % WARP_SIZE == 0);
+        constexpr int ThreadsDimY = THREAD_PER_GRID / WARP_SIZE;
+
+
+#define CALC_SRCB_SIZE(SrcT)   (0)
+        constexpr int SRCB_S_8  = CALC_SRCB_SIZE(uint8_t);
+        constexpr int SRCB_S_16 = CALC_SRCB_SIZE(uint16_t);
+        constexpr int SRCB_S_32 = CALC_SRCB_SIZE(uint32_t);
+#undef CALC_SRCB_SIZE
+        {   using SrcT = uint8_t; using DstT = uint8_t;
+            cudaFuncSetAttribute(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_8, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_8);
+        } if (!is_same<ValT, uint8_t>::value) { using SrcT = uint16_t; using DstT = uint8_t;
+            cudaFuncSetAttribute(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_16);
+        } if (!is_same<ValT, uint8_t>::value) { using SrcT = uint16_t; using DstT = uint16_t;
+            cudaFuncSetAttribute(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_16);
+        } if (is_same<ValT, uint32_t>::value) { using SrcT = uint32_t; using DstT = uint16_t;
+            cudaFuncSetAttribute(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_32);
+        } if (is_same<ValT, uint32_t>::value) { using SrcT = uint32_t; using DstT = uint32_t;
+            cudaFuncSetAttribute(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_32);
+        }
+
+        {   using SrcT = uint8_t; using DstT = uint8_t;
+            cudaFuncSetCacheConfig(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_8, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncCachePreferShared);
+        } if (!is_same<ValT, uint8_t>::value) { using SrcT = uint16_t; using DstT = uint8_t;
+            cudaFuncSetCacheConfig(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncCachePreferShared);
+        } if (!is_same<ValT, uint8_t>::value) { using SrcT = uint16_t; using DstT = uint16_t;
+            cudaFuncSetCacheConfig(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncCachePreferShared);
+        } if (is_same<ValT, uint32_t>::value) { using SrcT = uint32_t; using DstT = uint16_t;
+            cudaFuncSetCacheConfig(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncCachePreferShared);
+        } if (is_same<ValT, uint32_t>::value) { using SrcT = uint32_t; using DstT = uint32_t;
+            cudaFuncSetCacheConfig(&WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32, SrcT, DstT, BlockT, XYIdxT, XIdxT>, cudaFuncCachePreferShared);
+        }
+
+        const uint32_t nsum_pos = get_nsum_pos();
+
+        ValT *now_cu = src_cu + (CH_NUM == 1 ? 0ull : size * (size_t)CH_NUM);
+        ValT *nxt_cu = now_cu + size;
+        XYIdxT *nsum_buf_test = (XYIdxT*)(nxt_cu + size);
+        XYIdxT *nsum_buf_test2 = nsum_buf_test + nsum_scan_buf_len;
+        XIdxT *buf_idx = (XIdxT*)(nsum_buf_test2 + nsum_scan_buf_len);
+        XIdxT *nxt_idx = (XIdxT*)(buf_idx + size);
+
+
+        const int val_bit_len_m1 = val_bit_len - 1;
+        int h = val_bit_len_m1;
+        if (src_unpacked == true) {
+            if (src_cu_const != nullptr) {
+                printf("[Error!] not support. %s:%d\n", __FILE__, __LINE__);
+                exit(-1);
+            }
+            WaveletMatrix2dCu5C_first_gpu_multi_srcunpacked<ThreadsDimY> <<<grid, thread, 0, main_stream >>> (ValT(mask / 2),block_pair_num, size_div_warp, now_cu, nsum_buf_test, buf_byte_div32, buf_idx, W, (XYIdxT)W * H);
+        } else if (src_cu_const == nullptr) {
+            WaveletMatrix2dCu5C_first_gpu_multi<CH_NUM, ThreadsDimY, 0> <<<grid_x, thread, 0, main_stream >>> (ValT(mask / 2),block_pair_num, size_div_warp, src_cu, now_cu, nsum_buf_test, buf_byte_div32, buf_idx, W, (XYIdxT)W * H, src_cu_const);
+        } else {
+            WaveletMatrix2dCu5C_first_gpu_multi<CH_NUM, ThreadsDimY, 1> <<<grid_x, thread, 0, main_stream >>> (ValT(mask / 2),block_pair_num, size_div_warp, src_cu, now_cu, nsum_buf_test, buf_byte_div32, buf_idx, W, (XYIdxT)W * H, src_cu_const);
+        }
+        BlockT * nsum_p = get_bv_block_cu(h) + nsum_pos;
+        WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test, nsum_buf_test2, nsum_p, buf_byte_div32, bv_block_byte_div32);
+
+        const XYIdxT cwm_buf_byte_div32 = WM.get_buf_byte_div32();
+
+        if constexpr (sizeof(ValT) >= 4) for (; h > 16; --h) {
+            using SrcT = uint32_t;
+            using DstT = uint32_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            const int hp1 = std::min(val_bit_len - 1, h+1);
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(hp1 * CH_NUM);
+            XIdxT* cwm = (h + 1 == val_bit_len ? nullptr : WM.get_src_p(hp1 * CH_NUM));
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(hp1);
+            WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32> <<<grid, thread, SRCB_S_32, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_buf_test, nsum_buf_test2, bv_block_byte_div32, buf_byte_div32, buf_idx, inf, cwm, nxt_idx, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h);
+
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test2, nsum_buf_test, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_buf_test, nsum_buf_test2);
+            swap(now_cu, nxt_cu);
+            swap(buf_idx, nxt_idx);
+        }
+        if constexpr (sizeof(ValT) >= 4) if (h == 16 || (is_same<ValT, uint32_t>::value && h >= 0)) do {
+            using SrcT = uint32_t;
+            using DstT = uint16_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            const int hp1 = std::min(val_bit_len - 1, h+1);
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(hp1 * CH_NUM);
+            XIdxT* cwm = (h + 1 == val_bit_len ? nullptr : WM.get_src_p(hp1 * CH_NUM));
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(hp1);
+            WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_32> <<<grid, thread, SRCB_S_32, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_buf_test, nsum_buf_test2, bv_block_byte_div32, buf_byte_div32, buf_idx, inf, cwm, nxt_idx, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h);
+
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test2, nsum_buf_test, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_buf_test, nsum_buf_test2);
+            swap(now_cu, nxt_cu);
+            swap(buf_idx, nxt_idx);
+            --h;
+        } while(0);
+        if constexpr (sizeof(ValT) >= 2) for (; h > 8; --h) {
+            using SrcT = uint16_t;
+            using DstT = uint16_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            const int hp1 = std::min(val_bit_len - 1, h+1);
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(hp1 * CH_NUM);
+            XIdxT* cwm = (h + 1 == val_bit_len ? nullptr : WM.get_src_p(hp1 * CH_NUM));
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(hp1);
+            WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16> <<<grid, thread, SRCB_S_16, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_buf_test, nsum_buf_test2, bv_block_byte_div32, buf_byte_div32, buf_idx, inf, cwm, nxt_idx, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h);
+
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test2, nsum_buf_test, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_buf_test, nsum_buf_test2);
+            swap(now_cu, nxt_cu);
+            swap(buf_idx, nxt_idx);
+        }
+        if constexpr (sizeof(ValT) >= 2) if (h == 8 || (is_same<ValT, uint32_t>::value && h >= 0)) do {
+            using SrcT = uint16_t;
+            using DstT = uint8_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            const int hp1 = std::min(val_bit_len - 1, h+1);
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(hp1 * CH_NUM);
+            XIdxT* cwm = (h + 1 == val_bit_len ? nullptr : WM.get_src_p(hp1 * CH_NUM));
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(hp1);
+            WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_16> <<<grid, thread, SRCB_S_16, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_buf_test, nsum_buf_test2, bv_block_byte_div32, buf_byte_div32, buf_idx, inf, cwm, nxt_idx, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h);
+
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test2, nsum_buf_test, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_buf_test, nsum_buf_test2);
+            swap(now_cu, nxt_cu);
+            swap(buf_idx, nxt_idx);
+            --h;
+        } while(0);
+        for (; h >= 0; --h) {
+            using SrcT = uint8_t;
+            using DstT = uint8_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            const int hp1 = std::min(val_bit_len - 1, h+1);
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(hp1 * CH_NUM);
+            XIdxT* cwm = (h + 1 == val_bit_len ? nullptr : WM.get_src_p(hp1 * CH_NUM));
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(hp1);
+            WaveletMatrix2dCu5C_UpSweep_gpu<CH_NUM, ThreadsDimY, SRC_CACHE_DIV, SRCB_S_8> <<<grid, thread, SRCB_S_8, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_buf_test, nsum_buf_test2, bv_block_byte_div32, buf_byte_div32, buf_idx, inf, cwm, nxt_idx, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h);
+
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrix2dCu5C_ExclusiveSum<CH_NUM, MAX_BLOCK_X, XYIdxT> <<< wm_num, grid_x, 0, main_stream >>> (nsum_buf_test2, nsum_buf_test, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_buf_test, nsum_buf_test2);
+            swap(now_cu, nxt_cu);
+            swap(buf_idx, nxt_idx);
+        }
+        {
+            const int h = 0;
+            XYIdxT* WM_nsum_buf = WM.get_nsum_buf(h * CH_NUM);
+            XIdxT* cwm = WM.get_src_p(h * CH_NUM);
+            BlockT* bv_block_nbit_pre_cu_h = get_bv_block_cu(h);
+
+            WaveletMatrix2dCu5C_last_gpu<CH_NUM, ThreadsDimY, BlockT> <<<grid, thread, SRCB_S_8, main_stream >>> (block_pair_num, size_div_w, buf_byte_div32, nxt_idx, inf, cwm, WM_nsum_buf, cwm_buf_byte_div32, bv_block_nbit_pre_cu_h, bv_block_byte_div32);
+        }
+        WM.construct(main_stream, false);
+    }
+
+    XYIdxT get_nsum_pos() const {
+        const XYIdxT size_div_w = size / WORD_SIZE;
+        return size_div_w;
+    }
+
+    template<int TH_W = 8, int TH_H = 32, typename ResTableT = std::nullptr_t, bool CUT_BORDER = false>
+    void median2d(const int r, const ResTableT* res_table = nullptr) {
+        median2d<TH_W, TH_H, ResTableT, CUT_BORDER>(r, -1, res_table);
+    }
+
+    template<int TH_W = 8, int TH_H = 32, typename ResTableT = std::nullptr_t, bool CUT_BORDER = false>
+    void median2d(const int r, int res_step_num = -1, const ResTableT* res_table = nullptr, const cudaStream_t main_stream = 0) {
+        if (bv_block_nbit_and_nsum_base_cu == nullptr) { printf("Median2d Error: memory not alloced."); return;}
+        if (is_same<ResTableT, std::nullptr_t>::value == false && res_table == nullptr) {printf("Median2d Error: res_table is null."); return;}
+        static_assert(is_same<ResTableT, std::nullptr_t>::value || (sizeof(ResTableT) <= sizeof(ValT)), "");
+
+        static_assert(TH_W * TH_H <= 1024, "max number of threads in block");
+
+        if (res_step_num < 0) res_step_num = W * CH_NUM;
+
+        constexpr int THREADS_NUM_W = TH_W;
+        const dim3 thread(THREADS_NUM_W, TH_H);
+        const dim3 grid(div_ceil<size_t>((CUT_BORDER ? W - 2 * r: W), THREADS_NUM_W), div_ceil<size_t>((CUT_BORDER ? H - 2 * r : H), TH_H), CH_NUM);
+
+
+        const uint32_t bv_nsum_pos = get_nsum_pos();
+        const BlockT*  bv_bv_block_nbit_cu_first = get_bv_block_cu(val_bit_len - 1);
+
+        const BlockT* wm_bv_block_nbit_cu_first = WM.get_bv_block_cu(w_bit_len - 1, (val_bit_len - 1) * CH_NUM); //
+        const uint32_t nsum_pos = WM.get_nsum_pos();
+        const uint64_t wm_bv_block_byte = WM.get_bv_block_byte();
+
+        if (bv_nsum_pos != nsum_pos) { printf("err! line %d", __LINE__); exit(-1); }
+        if (WM.get_bv_block_byte()  != WM.get_bv_block_h_byte_div32() * 32ull * w_bit_len) { printf("err! line %d", __LINE__); exit(-1); }
+
+        if (bv_block_len != WM.bv_block_len) {printf("bv_block_len error!\n"); exit(1);}
+
+        using ResT = typename std::conditional<is_same<ResTableT, std::nullptr_t>::value, ValT, ResTableT>::type;
+
+        const int Wc = (CUT_BORDER ? W - 2 * r : W);
+        const int Hc = (CUT_BORDER ? H - 2 * r : H);
+
+        constexpr int VAL_BIT_LEN = (sizeof(ValT) < 4) ? MAX_BIT_LEN : -1;
+        WaveletMatrix2dCu5C_median2d_cu<CH_NUM, THREADS_NUM_W, TH_H, VAL_BIT_LEN, CUT_BORDER, XYIdxT, XIdxT, ValT> <<<grid, thread, 0, main_stream>>>
+            (Hc, Wc, res_step_num, r, (ResT*)res_cu, wm_bv_block_nbit_cu_first, nsum_pos, WM.get_bv_block_h_byte_div32(), bv_block_len,
+                bv_bv_block_nbit_cu_first, w_bit_len, val_bit_len, res_table);
+    }
+
+    template<typename ResTableT = ValT>
+    vector<vector<ResTableT>> get_res() {
+        static_assert(sizeof(ResTableT) <= sizeof(ValT), "");
+        auto res = vector<vector<ResTableT>>(H, vector<ResTableT>(W));
+        if (res_cu == nullptr) { printf("get_res Error: memory not alloced."); return res;}
+
+        for (int i = 0; i < H; ++i) {
+            cudaMemcpy(res[i].data(), res_cu + (XYIdxT)W * i, W * sizeof(ResTableT), cudaMemcpyDeviceToHost);
+        }
+        return res;
+    }
+};
+
+} // end namespace wavelet_median
+}}} //end namespace cv::cuda::device
+
+#endif
+#endif // __OPENCV_WAVELET_MATRIX_2D_CUH__
diff --git a/modules/cudafilters/src/cuda/wavelet_matrix_feature_support_checks.h b/modules/cudafilters/src/cuda/wavelet_matrix_feature_support_checks.h
new file mode 100644
index 00000000000..6c47cc5e1a4
--- /dev/null
+++ b/modules/cudafilters/src/cuda/wavelet_matrix_feature_support_checks.h
@@ -0,0 +1,82 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_WAVELET_MATRIX_FEATURE_SUPPORT_CHECKS_H__
+#define __OPENCV_WAVELET_MATRIX_FEATURE_SUPPORT_CHECKS_H__
+
+#ifdef HAVE_CUDA
+#include <cuda.h>
+#include <cuda_runtime_api.h>
+#endif // HAVE_CUDA
+
+
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#if CUDA_VERSION >= 11000 || CUDART_VERSION >= 11000
+
+
+// Check `if constexpr` is available.
+
+// GCC has been supported since 7.1
+#if defined(__GNUC__) && (__GNUC__ > 7 || (__GNUC__ == 7 && __GNUC_MINOR__ >= 1))
+#define __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+#endif
+
+// clang has been supported since 5.0
+#if defined(__clang__) && (__clang_major__ >= 5)
+#define __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+#endif
+
+
+// Visual Studio has been supported since Visual Studio 2019 (16.1.2)
+#if defined(_MSC_VER) && _MSC_VER >= 1921
+#define __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+#endif
+
+
+// I confirmed that it works with Intel C++ Compiler 2021.1.2. It did not work with icc 19.0.1.
+#if defined(__INTEL_COMPILER) && (__INTEL_COMPILER > 202101 || (__INTEL_COMPILER == 202101 && __INTEL_COMPILER_UPDATE >= 2))
+#define __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+#endif
+
+#endif // CUDA_VERSION
+#endif // __OPENCV_WAVELET_MATRIX_FEATURE_SUPPORT_CHECKS_H__
diff --git a/modules/cudafilters/src/cuda/wavelet_matrix_float_supporter.cuh b/modules/cudafilters/src/cuda/wavelet_matrix_float_supporter.cuh
new file mode 100644
index 00000000000..3a21a3c7082
--- /dev/null
+++ b/modules/cudafilters/src/cuda/wavelet_matrix_float_supporter.cuh
@@ -0,0 +1,227 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_WAVELET_MATRIX_FLOAT_SUPPORTER_CUH__
+#define __OPENCV_WAVELET_MATRIX_FLOAT_SUPPORTER_CUH__
+
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#include "wavelet_matrix_feature_support_checks.h"
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+
+
+namespace cv { namespace cuda { namespace device
+{
+
+namespace wavelet_matrix_median {
+namespace WMMedianFloatSupporter {
+
+template<int blockDim, typename IdxT>
+__global__ void iota_idx1(IdxT *idx_in_cu, const IdxT hw) {
+    const IdxT i = blockIdx.x * blockDim + threadIdx.x;
+    if (i >= hw) return;
+    idx_in_cu[i] = i;
+}
+
+template<int blockDim, int CH_NUM, typename IdxT, typename ValT>
+__global__ void split_and_iota_idx(IdxT *idx_in_cu, const ValT* val_in_cu, ValT* val_out_cu, const IdxT hw) {
+    const size_t i = blockIdx.x * blockDim + threadIdx.x;
+    if (i >= hw) return;
+
+    static_assert(is_same<ValT, float>::value, "");
+    // static_assert(2 <= CH_NUM && CH_NUM <= 4);
+    using SrcTU = std::conditional_t<CH_NUM == 2, float2, conditional_t<CH_NUM == 3, float3, float4>>;
+
+    const SrcTU *src_u = (SrcTU*)val_in_cu;
+    const SrcTU src_uv = src_u[i];
+
+    if (CH_NUM >= 1) { // constexpr
+        val_out_cu[i] = src_uv.x;
+        idx_in_cu[i] = i;
+    }
+    if constexpr (CH_NUM >= 2) {
+        val_out_cu += hw; idx_in_cu += hw;
+        val_out_cu[i] = src_uv.y;
+        idx_in_cu[i] = i;
+    }
+    if constexpr (CH_NUM >= 3) {
+        val_out_cu += hw; idx_in_cu += hw;
+        val_out_cu[i] = src_uv.z;
+        idx_in_cu[i] = i;
+    }
+    if constexpr (CH_NUM >= 4) {
+        val_out_cu += hw; idx_in_cu += hw;
+        val_out_cu[i] = src_uv.w;
+        idx_in_cu[i] = i;
+    }
+}
+
+template<int blockDim, typename IdxT>
+__global__ void set_wm_val_1(IdxT *wm_src_p, const IdxT *idx_out_cu, const IdxT hw) {
+    const IdxT i = blockIdx.x * blockDim + threadIdx.x;
+    if (i >= hw) return;
+    const IdxT j = idx_out_cu[i];
+    wm_src_p[j] = i;
+}
+
+template<int blockDim, int CH_NUM, typename IdxT>
+__global__ void set_wm_val(IdxT *wm_src_p, const IdxT *idx_out_cu, const IdxT hw, const IdxT buf_byte_div32) {
+    const IdxT i = blockIdx.x * blockDim + threadIdx.x;
+    if (i >= hw) return;
+    const size_t hwc = size_t(hw) * blockIdx.y;
+    const IdxT j = idx_out_cu[i + hwc];
+    const size_t src_offset = buf_byte_div32 * 32ull / sizeof(IdxT) * blockIdx.y;
+    wm_src_p[src_offset + j] = i;
+}
+
+template<int blockDim, typename IdxT, typename ValT>
+__global__ void conv_res_cu(ValT *dst, const ValT *val_out_cu, const IdxT *res_cu, const IdxT hw) {
+    const IdxT i = blockIdx.x * blockDim + threadIdx.x;
+    if (i >= hw) return;
+
+    const IdxT r = res_cu[i];
+    dst[i] = val_out_cu[r];
+}
+
+template<typename ValT, int CH_NUM, typename IdxT>
+struct WMMedianFloatSupporter {
+    constexpr static int blockDim = 512;
+    int h = 0, w = 0;
+    int hw_bit_len = -1;
+    WMMedianFloatSupporter(){};
+    WMMedianFloatSupporter(int h, int w) { reset(h, w); }
+    ~WMMedianFloatSupporter(){
+        free();
+    }
+    ValT *val_in_cu = nullptr;
+    IdxT *idx_in_cu = nullptr;
+    ValT *val_out_cu = nullptr;
+    IdxT *idx_out_cu = nullptr;
+    void *cub_temp_storage = nullptr;
+    size_t cub_temp_storage_bytes;
+    // set: val_in
+    // get: val_out
+    //  1ch
+    //  [val_in][idx_in][val_out][idx_out][cub_temp]
+    //  C ch
+
+    //  [val_in][......][......][......]
+    //  AaBbCcDd
+    //  [^^^^^^][......][valin2][idxin2]
+    //                  ABCDabcd01230123
+
+    //  [val0in][val1in][idx0in][idx1in][val0out][val1out][idx0out][idx1out][cub_temp][d_offsets]
+
+    void reset(const int H, const int W) {
+        h = H; w = W;
+        free();
+    }
+    void alloc(){
+        const size_t hwc = size_t(CH_NUM) * h * w;
+        if (CH_NUM == 1) { // constexpr
+            cub::DeviceRadixSort::SortPairs(
+                nullptr, cub_temp_storage_bytes, val_in_cu, val_out_cu, idx_in_cu, idx_out_cu, hwc);
+            cudaMalloc(&val_in_cu, 2ull * hwc * (sizeof(ValT) + sizeof(IdxT)) + cub_temp_storage_bytes);
+        } else {
+            cub::DeviceSegmentedRadixSort::SortPairs(
+                nullptr, cub_temp_storage_bytes, val_in_cu, val_out_cu, idx_in_cu, idx_out_cu, hwc, CH_NUM, (int*)nullptr, (int*)nullptr);
+            const size_t offsets_arr_size = (CH_NUM + 1) * sizeof(int);
+            cudaMalloc(&val_in_cu, 2ull * hwc * (sizeof(ValT) + sizeof(IdxT)) + cub_temp_storage_bytes + offsets_arr_size);
+        }
+        idx_in_cu = (IdxT*)(val_in_cu + hwc);
+        val_out_cu = (ValT*)(idx_in_cu + hwc);
+        idx_out_cu = (IdxT*)(val_out_cu + hwc);
+        int *d_offsets = (int*)(idx_out_cu + hwc);
+        cub_temp_storage = d_offsets + (CH_NUM + 1);
+    }
+    void free() {
+        if (val_in_cu != nullptr) {
+            cudaFree(val_in_cu);
+        }
+    }
+    void sort_and_set(IdxT *wm_src_p, const IdxT buf_byte_div32 = 0){
+        const IdxT hw = h * w;
+        const size_t hwc = size_t(CH_NUM) * hw;
+        const dim3 gridDim((hw + blockDim - 1) / blockDim, CH_NUM);
+
+        if (CH_NUM == 1) { // constexpr
+            iota_idx1<blockDim><<<gridDim, blockDim>>>(idx_in_cu, hw);
+            cub::DeviceRadixSort::SortPairs(
+                cub_temp_storage, cub_temp_storage_bytes, val_in_cu, val_out_cu, idx_in_cu, idx_out_cu, hw);
+            set_wm_val_1<blockDim><<<gridDim, blockDim>>>(wm_src_p, idx_out_cu, hw);
+        } else {
+            auto idx2 = idx_out_cu;
+            auto val2 = val_out_cu;
+            auto idx3 = idx_in_cu;
+            auto val3 = val_in_cu;
+            split_and_iota_idx<blockDim, CH_NUM><<<gridDim.x, blockDim>>>(idx2, val_in_cu, val2, hw);
+
+            int h_offsets[CH_NUM + 1];
+            for (size_t i = 0; i <= CH_NUM; ++i) h_offsets[i] = i * hw;
+            int *d_offsets = (int*)(idx_out_cu + hwc);
+            cudaMemcpy(d_offsets, h_offsets, (CH_NUM + 1) * sizeof(int), cudaMemcpyHostToDevice);
+
+
+            cub::DeviceSegmentedRadixSort::SortPairs(
+                cub_temp_storage, cub_temp_storage_bytes, val2, val3, idx2, idx3, hwc, CH_NUM, d_offsets, d_offsets + 1);
+            set_wm_val<blockDim, CH_NUM><<<gridDim, blockDim>>>(wm_src_p, idx3, hw, buf_byte_div32);
+        }
+        for(hw_bit_len = 1; ; ++hw_bit_len) {
+            if ((1ull << hw_bit_len) >= hw) {
+                break;
+            }
+        }
+    }
+    const ValT* get_res_table() const {
+        if (CH_NUM == 1) { // constexpr
+            return val_out_cu;
+        } else {
+            return val_in_cu;
+        }
+    }
+};
+} // end namespace WMMedianFloatSupporter
+} // end namespace wavelet_matrix_median
+
+}}} //end namespace cv::cuda::device
+#endif
+#endif // __OPENCV_WAVELET_MATRIX_FLOAT_SUPPORTER_CUH__
diff --git a/modules/cudafilters/src/cuda/wavelet_matrix_multi.cuh b/modules/cudafilters/src/cuda/wavelet_matrix_multi.cuh
new file mode 100644
index 00000000000..de59d75993a
--- /dev/null
+++ b/modules/cudafilters/src/cuda/wavelet_matrix_multi.cuh
@@ -0,0 +1,636 @@
+/*M///////////////////////////////////////////////////////////////////////////////////////
+//
+//  IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
+//
+//  By downloading, copying, installing or using the software you agree to this license.
+//  If you do not agree to this license, do not download, install,
+//  copy or use the software.
+//
+//
+//                           License Agreement
+//                For Open Source Computer Vision Library
+//
+// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
+// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
+// Third party copyrights are property of their respective owners.
+//
+// Redistribution and use in source and binary forms, with or without modification,
+// are permitted provided that the following conditions are met:
+//
+//   * Redistribution's of source code must retain the above copyright notice,
+//     this list of conditions and the following disclaimer.
+//
+//   * Redistribution's in binary form must reproduce the above copyright notice,
+//     this list of conditions and the following disclaimer in the documentation
+//     and/or other materials provided with the distribution.
+//
+//   * The name of the copyright holders may not be used to endorse or promote products
+//     derived from this software without specific prior written permission.
+//
+// This software is provided by the copyright holders and contributors "as is" and
+// any express or implied warranties, including, but not limited to, the implied
+// warranties of merchantability and fitness for a particular purpose are disclaimed.
+// In no event shall the Intel Corporation or contributors be liable for any direct,
+// indirect, incidental, special, exemplary, or consequential damages
+// (including, but not limited to, procurement of substitute goods or services;
+// loss of use, data, or profits; or business interruption) however caused
+// and on any theory of liability, whether in contract, strict liability,
+// or tort (including negligence or otherwise) arising in any way out of
+// the use of this software, even if advised of the possibility of such damage.
+//
+//M*/
+
+#ifndef __OPENCV_WAVELET_MATRIX_MULTI_CUH__
+#define __OPENCV_WAVELET_MATRIX_MULTI_CUH__
+
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#include "wavelet_matrix_feature_support_checks.h"
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+
+
+#include <cuda_runtime.h>
+#include <device_launch_parameters.h>
+#include <cub/cub.cuh>
+#include <cub/device/device_scan.cuh>
+#include <cub/block/block_scan.cuh>
+#include "opencv2/core/cuda/warp_shuffle.hpp"
+
+
+#include <vector>
+
+namespace cv { namespace cuda { namespace device
+{
+
+namespace wavelet_matrix_median {
+    using std::vector;
+    using namespace std;
+
+    template <typename T>
+    constexpr T div_ceil(T a, T b) {
+        return (a + b - 1) / b;
+    }
+
+
+template<int ThreadsDimY, int SRCB_S, int DSTB_S, typename SrcT, typename DstT, class BlockT, typename IdxType>
+__global__ void WaveletMatrixMultiCu4G_UpSweep_gpu(const SrcT mask, const uint16_t block_pair_num, const IdxType size_div_w, const SrcT* __restrict__ src, DstT* __restrict__ dst, BlockT* __restrict__ nbit_bp, const IdxType* __restrict__ nsum_zeros_buf, IdxType* __restrict__ nsum_zeros_buf2, const uint32_t bv_block_byte_div32, const uint32_t buf_byte_div32) {
+    using WordT = decltype(BlockT::nbit);
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+    constexpr int WORD_SIZE = 8 * sizeof(WordT);
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "");
+    constexpr uint32_t WORD_DIV_WARP = WORD_SIZE / WARP_SIZE;
+
+    src            = (SrcT*)((uint8_t*)src + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+    dst            = (DstT*)((uint8_t*)dst + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+    nsum_zeros_buf  = (IdxType*)((uint8_t*)nsum_zeros_buf + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+    nsum_zeros_buf2 = (IdxType*)((uint8_t*)nsum_zeros_buf2 + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+    nbit_bp        = (BlockT*)((uint8_t*)nbit_bp + (size_t)blockIdx.y * (bv_block_byte_div32*32ull)); // TODO: rename
+
+    using WarpScan = cub::WarpScan<IdxType>;
+    using WarpScanY = cub::WarpScan<IdxType, ThreadsDimY>;
+    using WarpReduce = cub::WarpReduce<uint32_t>;
+    using WarpReduceY = cub::WarpReduce<uint32_t, ThreadsDimY>;
+
+    constexpr size_t shmem_size = sizeof(SrcT) * (ThreadsDimY * (WARP_SIZE - 1) * WARP_SIZE);
+    static_assert(SRCB_S == shmem_size, "");
+    static_assert(SRCB_S + DSTB_S < 64 * 1024, "");
+
+    constexpr int DST_BUF_SIZE = DSTB_S;
+    constexpr int DST_BUF_NUM_PER_WARP = DST_BUF_SIZE / (ThreadsDimY * sizeof(DstT)); // [32k/32/2=512] [48k/8/1=6114]
+    constexpr int DST_BUF_NUM_PER_THREAD = DST_BUF_NUM_PER_WARP / WARP_SIZE;
+    static_assert(DST_BUF_NUM_PER_THREAD <= WARP_SIZE, "");
+
+
+    extern __shared__ uint8_t shmem_base[];
+    SrcT* __restrict__ src_val_cache = (SrcT*)shmem_base;
+    DstT* __restrict__ dst_buf = (DstT*)&src_val_cache[SRCB_S] + threadIdx.y * DST_BUF_NUM_PER_WARP;  //[ThreadsDimY][DST_BUF_NUM_PER_WARP];
+
+    __shared__ uint4 nsum_count_sh[ThreadsDimY];
+    __shared__ IdxType pre_sum_share[2];
+    __shared__ IdxType warp_scan_sums[WARP_SIZE];
+    __shared__ typename WarpScan::TempStorage s_scanStorage;
+    __shared__ typename WarpScanY::TempStorage s_scanStorage2;
+    __shared__ typename WarpReduce::TempStorage WarpReduce_temp_storage[ThreadsDimY];
+    __shared__ typename WarpReduceY::TempStorage WarpReduceY_temp_storage;
+    // shmem ------ end ------
+
+    const IdxType size_div_warp = size_div_w * WORD_DIV_WARP;
+    const IdxType nsum = nbit_bp[size_div_w].nsum;
+    const IdxType nsum_offset = nsum_zeros_buf[blockIdx.x];
+
+
+    IdxType nsum_idx0_org = nsum_offset;
+    IdxType nsum_idx1_org = (IdxType)blockIdx.x * block_pair_num * THREAD_PER_GRID + nsum - nsum_idx0_org;
+    nsum_idx0_org /= (IdxType)block_pair_num * ThreadsDimY * WARP_SIZE;
+    nsum_idx1_org /= (IdxType)block_pair_num * ThreadsDimY * WARP_SIZE;
+    const IdxType nsum_idx0_bound = (nsum_idx0_org + 1) * block_pair_num * ThreadsDimY * WARP_SIZE;
+    const IdxType nsum_idx1_bound = (nsum_idx1_org + 1) * block_pair_num * ThreadsDimY * WARP_SIZE;
+    uint4 nsum_count = make_uint4(0, 0, 0, 0);
+
+    const unsigned short th_idx = threadIdx.y * WARP_SIZE + threadIdx.x;
+    if (th_idx == 0) {
+        pre_sum_share[0] = nsum_offset;
+    }
+
+
+    for (IdxType ka = 0; ka < block_pair_num; ka += WARP_SIZE) {
+        const IdxType ibb = ((IdxType)blockIdx.x * block_pair_num + ka) * ThreadsDimY;
+        if (ibb >= size_div_warp) break;
+        WarpWT my_bits = 0;
+        SrcT first_val;
+
+        const IdxType src_val_cache_offset = IdxType(threadIdx.y * (WARP_SIZE - 1) - 1) * WARP_SIZE + threadIdx.x;
+        for (IdxType kb = 0, i = ibb + WARP_SIZE * threadIdx.y; kb < WARP_SIZE; ++kb, ++i) {
+            if (i >= size_div_warp) break;
+            WarpWT bits;
+            const SrcT v = src[i * WARP_SIZE + threadIdx.x];
+            if (kb == 0) {
+                first_val = v;
+            } else {
+                src_val_cache[src_val_cache_offset + kb * WARP_SIZE] = v;
+            }
+            if (v <= mask) {
+                bits = __activemask();
+            } else {
+                bits = ~__activemask();
+            }
+            if (threadIdx.x == kb) {
+                my_bits = bits;
+            }
+        }
+        IdxType t, c = __popc(my_bits);
+        WarpScan(s_scanStorage).ExclusiveSum(c, t);
+
+        if (threadIdx.x == WARP_SIZE - 1) {
+            warp_scan_sums[threadIdx.y] = c + t;
+        }
+        __syncthreads();
+        IdxType pre_sum = pre_sum_share[(ka & WARP_SIZE) > 0];
+        IdxType s = threadIdx.x < ThreadsDimY ? warp_scan_sums[threadIdx.x] : 0;
+        WarpScanY(s_scanStorage2).ExclusiveSum(s, s);
+
+        s = cv::cuda::device::shfl(s, threadIdx.y, WARP_SIZE);
+
+        s += t + pre_sum;
+        if (th_idx == THREAD_PER_GRID - 1) {
+            pre_sum_share[(ka & WARP_SIZE) == 0] = s + c;
+        }
+        const IdxType bi = ibb + th_idx;
+
+        if (bi < size_div_warp) {
+            static_assert(WORD_SIZE == 32, "");
+            nbit_bp[bi] = BlockT{s, my_bits};
+        }
+        if (mask == 0) continue;
+
+        const SrcT mask_2 = mask >> 1;
+        SrcT vo = first_val;
+        for (IdxType j = 0, i = ibb + WARP_SIZE * threadIdx.y; j < WARP_SIZE;) {
+            if (i >= size_div_warp) break;
+
+
+            IdxType idx0_begin, idx0_num, idx1_offset, idx01_num, idx1_offset0;
+            if (DST_BUF_SIZE > 0) { // constexpr
+                IdxType idx1_begin, idx0_end;
+                const IdxType ib = ::min(size_div_warp, i + DST_BUF_NUM_PER_THREAD);
+                const IdxType jb = j + ib - i - 1;
+                idx0_begin = cv::cuda::device::shfl(s, j, WARP_SIZE);
+                idx1_begin = i * WARP_SIZE + nsum - idx0_begin;
+                idx0_end = cv::cuda::device::shfl(s + c, jb, WARP_SIZE);
+
+                idx0_num = idx0_end - idx0_begin;
+                idx1_offset = idx1_begin - idx0_num;
+                idx01_num = (ib - i) * WARP_SIZE;
+                idx1_offset0 = nsum - idx1_begin + idx0_num;
+            }
+            constexpr int DST_LOOP_NUM = (DST_BUF_SIZE == 0 ? 1 : DST_BUF_NUM_PER_THREAD);
+            for (IdxType kb = 0; kb < DST_LOOP_NUM; ++kb, ++j, ++i) {
+                if (i >= size_div_warp) break;
+
+                const WarpWT  e_nbit = cv::cuda::device::shfl(my_bits, j, WARP_SIZE);
+                const IdxType e_nsum = cv::cuda::device::shfl(s, j, WARP_SIZE);
+                IdxType rank = __popc(e_nbit << (WARP_SIZE - threadIdx.x));
+                const IdxType idx0 = e_nsum + rank;
+
+                DstT v = (DstT)vo;
+                IdxType idx;
+                IdxType buf_idx;
+                if (vo > mask) { // 1
+                    const IdxType ij = i * WARP_SIZE + threadIdx.x;
+                    idx = ij + nsum - idx0;
+                    v &= mask;
+                    buf_idx = ij - idx0 + idx1_offset0;
+                } else {
+                    idx = idx0;
+                    buf_idx = idx0 - idx0_begin;
+                }
+                if (DST_BUF_SIZE == 0) {
+                    dst[idx] = v;
+                } else {
+                    dst_buf[buf_idx] = (DstT)v;
+                }
+
+                if (v <= mask_2) {
+                    if (vo <= mask) {
+                        if (idx < nsum_idx0_bound) {
+                            nsum_count.x++;
+                        } else {
+                            assert(idx < nsum_idx0_bound + block_pair_num * ThreadsDimY * WARP_SIZE);
+                            nsum_count.y++;
+                        }
+                    } else {
+                        if (idx < nsum_idx1_bound) {
+                            nsum_count.z++;
+                        } else {
+                            assert(idx < nsum_idx1_bound + block_pair_num * ThreadsDimY * WARP_SIZE);
+                            nsum_count.w++;
+                        }
+                    }
+                }
+                if (j == WARP_SIZE - 1) { j = WARP_SIZE; break; }
+                vo = src_val_cache[(threadIdx.y * (WARP_SIZE - 1) + j) * WARP_SIZE + threadIdx.x];
+            }
+            if (DST_BUF_SIZE > 0) { // constexpr
+                for (IdxType j = threadIdx.x; (int)j < DST_BUF_NUM_PER_WARP; j += WARP_SIZE) {
+                    if (j >= idx01_num) break;
+                    IdxType idx;
+                    if (j < idx0_num) { // 0
+                        idx = j + idx0_begin;
+                    } else {    // 1
+                        idx = j + idx1_offset;
+                    }
+                    dst[idx] = dst_buf[j];
+                }
+            }
+        }
+    }
+
+    if (blockIdx.x == gridDim.x - 1 && th_idx == 0) {
+        nbit_bp[size_div_warp / WORD_DIV_WARP].nsum = nsum;
+    }
+    if (mask == 0) return;
+
+    nsum_count.x = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.x);
+    nsum_count.y = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.y);
+    nsum_count.z = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.z);
+    nsum_count.w = WarpReduce(WarpReduce_temp_storage[threadIdx.y]).Sum(nsum_count.w);
+    if (threadIdx.x == 0) {
+        nsum_count_sh[threadIdx.y] = nsum_count;
+    }
+    __syncthreads();
+
+
+    if (threadIdx.x < ThreadsDimY) {
+        nsum_count = nsum_count_sh[threadIdx.x];
+        nsum_count.x = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.x);
+        nsum_count.y = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.y);
+        nsum_count.z = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.z);
+        nsum_count.w = WarpReduceY(WarpReduceY_temp_storage).Sum(nsum_count.w);
+        if (th_idx == 0) {
+            const IdxType nsum_idx0_org = nsum_idx0_bound / ((IdxType)block_pair_num * ThreadsDimY * WARP_SIZE);
+            const IdxType nsum_idx1_org = nsum_idx1_bound / ((IdxType)block_pair_num * ThreadsDimY * WARP_SIZE);
+            if (nsum_count.x > 0) atomicAdd(nsum_zeros_buf2 + nsum_idx0_org - 1, nsum_count.x);
+            if (nsum_count.y > 0) atomicAdd(nsum_zeros_buf2 + nsum_idx0_org - 0, nsum_count.y);
+            if (nsum_count.z > 0) atomicAdd(nsum_zeros_buf2 + nsum_idx1_org - 1, nsum_count.z);
+            if (nsum_count.w > 0) atomicAdd(nsum_zeros_buf2 + nsum_idx1_org - 0, nsum_count.w);
+        }
+    }
+}
+
+
+template<int MAX_BLOCK_X, typename IdxType, typename BlockT>
+__global__ void WaveletMatrixMultiCu4G_ExclusiveSum(IdxType* __restrict__ nsum_scan_buf, IdxType* __restrict__ nsum_zeros_buf2, BlockT* __restrict__ nsum_p, const uint32_t buf_byte_div32, const uint32_t bv_block_byte_div32) {
+
+    typedef cub::BlockScan<IdxType, MAX_BLOCK_X> BlockScan;
+    __shared__ typename BlockScan::TempStorage temp_storage;
+
+    IdxType thread_data1;
+    IdxType thread_data2;
+    nsum_scan_buf  = (IdxType*)((uint8_t*)nsum_scan_buf + (size_t)blockIdx.x * (buf_byte_div32*32ull));
+    nsum_zeros_buf2 = (IdxType*)((uint8_t*)nsum_zeros_buf2 + (size_t)blockIdx.x * (buf_byte_div32*32ull));
+
+    thread_data1 = nsum_scan_buf[threadIdx.x];
+    BlockScan(temp_storage).ExclusiveSum(thread_data1, thread_data2);
+
+    nsum_scan_buf[threadIdx.x] = thread_data2;
+    nsum_zeros_buf2[threadIdx.x] = 0;
+
+    if (threadIdx.x == blockDim.x - 1) {
+        thread_data2 += thread_data1;
+        nsum_p = (BlockT*)((uint8_t*)nsum_p + (size_t)blockIdx.x * (bv_block_byte_div32*32ull));
+        nsum_p->nsum = thread_data2;
+    }
+}
+
+
+template<int ThreadsDimY, typename SrcT, typename IdxType>
+__global__ void WaveletMatrixMultiCu4G_first_gpu(const SrcT mask, uint16_t block_pair_num, const IdxType size_div_warp, const SrcT* __restrict__ src, IdxType* __restrict__ nsum_scan_buf, const uint32_t buf_byte_div32) {
+    using WarpWT = uint32_t;
+    constexpr int WARP_SIZE = 8 * sizeof(WarpWT);
+    static_assert(WARP_SIZE == 32, "");
+    static constexpr int THREAD_PER_GRID = ThreadsDimY * WARP_SIZE;
+
+    src = (SrcT*)((uint8_t*)src + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+    IdxType cs = 0;
+    IdxType ibb = (IdxType)blockIdx.x * block_pair_num * ThreadsDimY;
+    for (IdxType ka = 0; ka < block_pair_num; ka += WARP_SIZE, ibb += THREAD_PER_GRID) {
+        for (IdxType kb = 0, i = ibb + WARP_SIZE * threadIdx.y; kb < WARP_SIZE; ++kb, ++i) {
+            if (i >= size_div_warp) break;
+            const SrcT v = src[i * WARP_SIZE + threadIdx.x];
+            if (v <= mask) {
+                ++cs;
+            }
+        }
+    }
+    using BlockReduce = cub::BlockReduce<IdxType, WARP_SIZE, cub::BLOCK_REDUCE_WARP_REDUCTIONS, ThreadsDimY>;
+    __shared__ typename BlockReduce::TempStorage s_reduceStorage;
+    IdxType reducedValue = BlockReduce(s_reduceStorage).Sum(cs);
+
+    if (threadIdx.y == 0 && threadIdx.x == 0) {
+        nsum_scan_buf = (IdxType*)((uint8_t*)nsum_scan_buf + (size_t)blockIdx.y * (buf_byte_div32*32ull));
+        nsum_scan_buf[blockIdx.x] = reducedValue;
+    }
+}
+
+
+
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_Z(uint32_t r) {
+    return (r >= MIN_DSTBUF_KB) ? r * 1024 : 0;
+}
+
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_B(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_Z<MIN_DSTBUF_KB>((r + 1) / 2);
+}
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A16(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_B<MIN_DSTBUF_KB>(r | (r >> 16));
+}
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A8(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A16<MIN_DSTBUF_KB>(r | (r >> 8));
+}
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A4(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A8<MIN_DSTBUF_KB>(r | (r >> 4));
+}
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A2(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A4<MIN_DSTBUF_KB>(r | (r >> 2));
+}
+template<int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A1(uint32_t r) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A2<MIN_DSTBUF_KB>(r | (r >> 1));
+}
+template<int SHMEM_USE_KB, int MIN_DSTBUF_KB>
+constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb(int SRCB_S) {
+    return WaveletMatrixMultiCu4G_get_dstbuf_kb_internal_A1<MIN_DSTBUF_KB>(SHMEM_USE_KB - (SRCB_S + 1023) / 1024);
+}
+
+// template<int SHMEM_USE_KB, int MIN_DSTBUF_KB>
+// constexpr int WaveletMatrixMultiCu4G_get_dstbuf_kb(int SRCB_S) {
+//     uint32_t r = (SHMEM_USE_KB - (SRCB_S + 1023) / 1024);
+//     r |= r >> 1;
+//     r |= r >> 2;
+//     r |= r >> 4;
+//     r |= r >> 8;
+//     r |= r >> 16;
+//     r = (r + 1) / 2;
+//     return (r >= MIN_DSTBUF_KB) ? r * 1024 : 0;
+// }
+
+
+template <typename T = uint64_t, int TH_NUM = 512, int WORD_SIZE = 32, typename IdxType = uint32_t>
+struct WaveletMatrixMultiCu4G {
+    static constexpr int MAX_BIT_LEN = 8 * sizeof(T);
+
+    static constexpr uint32_t WSIZE = WORD_SIZE;
+    using T_Type = T;
+    static constexpr int WARP_SIZE       = 32;
+    static constexpr int THREAD_PER_GRID = TH_NUM;
+    static constexpr int MAX_BLOCK_X     = 1024;
+    static_assert(WORD_SIZE == 32 || WORD_SIZE == 64, "WORD SIZE must be 32 or 64");
+    using WordT = typename std::conditional<WORD_SIZE == 32, uint32_t, uint64_t>::type;
+
+    static constexpr int SHMEM_USE_KB  = 64;
+
+    struct __align__(8) BLOCK32_T { uint32_t nsum; union { uint32_t nbit; uint32_t nbit_a[1];}; };
+    struct __align__(4) BLOCK64_T { uint32_t nsum; union { uint64_t nbit; uint32_t nbit_a[2];}; };
+    using BlockT = typename std::conditional<WORD_SIZE == 32, BLOCK32_T, BLOCK64_T>::type;
+
+    static constexpr int MIN_DSTBUF_KB =  4;
+    static constexpr int BLOCK_TYPE = 2;
+    using WarpWT = uint32_t;
+    static_assert(8 * sizeof(WarpWT) == WARP_SIZE, "bits of WarpWT must be WARP_SIZE");
+
+    IdxType size = 0;
+    int wm_num = 0;
+
+private:
+    T* src_cu = nullptr;
+    uint8_t* bv_block_nbit_and_nsum_base_cu = nullptr;
+    uint32_t bv_block_byte_div32;
+    uint32_t buf_byte_div32;
+public:
+    size_t bv_block_len = 0;
+    IdxType bv_zeros[MAX_BIT_LEN];
+    int bit_len = 0;
+
+    WaveletMatrixMultiCu4G(IdxType _n = 0, int _bit_len = 0, int num = 0) {
+        reset(_n, _bit_len, num);
+    }
+    void reset(IdxType _n, int _bit_len, int _num) {
+        cudaError_t err;
+        assert(size == 0 && src_cu == nullptr && 0 <= _bit_len && _bit_len <= MAX_BIT_LEN);
+        bit_len = _bit_len;
+        wm_num = _num;
+        if (_n == 0 || wm_num == 0) return;
+        size = div_ceil<size_t>(_n, WORD_SIZE) * WORD_SIZE;
+        bv_block_len = div_ceil<size_t>(size, THREAD_PER_GRID) * THREAD_PER_GRID / WORD_SIZE + 1;
+        bv_block_len = div_ceil<size_t>(bv_block_len, 8*2) * 8*2;
+
+        const size_t bv_block_byte = (sizeof(BlockT)) * bit_len * bv_block_len;
+        if (bv_block_byte % 32 != 0) { printf("bv_block_byte not 32n!"); exit(-1); }
+        bv_block_byte_div32 = div_ceil<size_t>(bv_block_byte, 32);
+
+        err = cudaMalloc(&bv_block_nbit_and_nsum_base_cu, (size_t)(bv_block_byte_div32*32ull) * _num);
+        if (bv_block_nbit_and_nsum_base_cu == nullptr) { printf("GPU Memory Alloc Error! %s:%d %s\n", __FILE__, __LINE__, cudaGetErrorString(err)); release(); return; }
+
+        const uint16_t block_pair_num = get_block_pair_num();
+        const IdxType nsum_scan_buf_len = div_ceil<size_t>(size, (size_t)THREAD_PER_GRID * block_pair_num);
+
+        const size_t buf_byte = sizeof(IdxType) * 2 * nsum_scan_buf_len + sizeof(T) * size * 2;
+        buf_byte_div32 = div_ceil<size_t>(buf_byte, 32);
+        err = cudaMalloc(&src_cu, (size_t)(buf_byte_div32*32ull) * _num);
+        if (src_cu == nullptr) { printf("GPU Memory Alloc Error! %s:%d %s\n", __FILE__, __LINE__, cudaGetErrorString(err)); release(); return; }
+
+    }
+    void release() {
+        size = 0;
+        if (src_cu != nullptr) cudaFree(src_cu);
+        if (bv_block_nbit_and_nsum_base_cu != nullptr) cudaFree(bv_block_nbit_and_nsum_base_cu);
+        src_cu = nullptr;
+        bv_block_nbit_and_nsum_base_cu = nullptr;
+    }
+    ~WaveletMatrixMultiCu4G() { release(); }
+
+    BlockT*  get_bv_block_cu(int h) const { return (BlockT*)(bv_block_nbit_and_nsum_base_cu + (sizeof(BlockT)) * bv_block_len * h); }
+    BlockT*  get_bv_block_cu(int h, int c) const { return (BlockT*)((uint8_t*)get_bv_block_cu(h) + (size_t)c * (bv_block_byte_div32*32ull)); }
+    uint64_t get_bv_block_byte() const { return (bv_block_byte_div32*32ull); }
+
+
+    T* get_src_p(int c) const { return src_cu + (size_t)(buf_byte_div32*32ull) / (sizeof(T)) * c; }
+
+    uint16_t get_block_pair_num() const {
+        constexpr int x_chunk = 65536 / THREAD_PER_GRID / WARP_SIZE; // To make the pixels assigned per grid a multiple of 65536.
+        static_assert(x_chunk > 0, "");
+        const uint64_t total_gridx = div_ceil<uint64_t>(size, THREAD_PER_GRID * WARP_SIZE);
+        uint64_t block_pair_num_org = div_ceil<uint64_t>(total_gridx, MAX_BLOCK_X);
+        if (block_pair_num_org <= x_chunk) {
+            block_pair_num_org--;
+            block_pair_num_org |= block_pair_num_org >> 1;
+            block_pair_num_org |= block_pair_num_org >> 2;
+            block_pair_num_org |= block_pair_num_org >> 4;
+            block_pair_num_org |= block_pair_num_org >> 8;
+            block_pair_num_org++;
+        } else {
+            block_pair_num_org = div_ceil<uint64_t>(block_pair_num_org, x_chunk) * x_chunk;
+        }
+        block_pair_num_org *= WARP_SIZE;
+
+        if (block_pair_num_org >= (1LL << (8 * sizeof(uint16_t)))) { printf("over block_pair_num %ld\n", block_pair_num_org); exit(1); }
+        return (uint16_t)block_pair_num_org;
+    }
+    std::pair<IdxType*, size_t> get_nsum_buf_and_buf_byte() const {
+        IdxType* nsum_buf = (IdxType*)(src_cu + 2ull * size);
+        return { nsum_buf, (buf_byte_div32*32ull) };
+    }
+    IdxType* get_nsum_buf(int c) const {
+        IdxType* nsum_buf = (IdxType*)(src_cu + 2ull * size);
+        return (IdxType*)((uint8_t*)nsum_buf + (size_t)(buf_byte_div32*32ull) * c);
+    }
+    uint64_t get_buf_byte() const { return (buf_byte_div32*32ull); }
+    uint32_t get_buf_byte_div32() const { return buf_byte_div32; }
+
+    IdxType get_nsum_scan_buf_len(uint16_t block_pair_num) const {
+        return div_ceil<IdxType>(size, THREAD_PER_GRID * block_pair_num);
+    }
+    IdxType get_nsum_scan_buf_len() const {
+        const uint16_t block_pair_num = get_block_pair_num();
+        return get_nsum_scan_buf_len(block_pair_num);
+    }
+
+    // Set data in src_cu before calling (data will be destroyed).
+    void construct(const cudaStream_t main_stream = 0, const bool run_first = true) {
+        assert(size > 0 && src_cu != nullptr);
+        if (size == 0 || wm_num == 0) return;
+        if (src_cu == nullptr) { printf("Build Error: memory not alloced."); return;}
+
+        T mask = ((T)1 << bit_len) - 1;
+
+        const uint16_t block_pair_num = get_block_pair_num();
+        const int grid_x = div_ceil<int>(size, THREAD_PER_GRID * block_pair_num);
+        if (grid_x > MAX_BLOCK_X) { printf("over grid_x %d\n", grid_x); exit(1); }
+
+
+        const dim3 grid(grid_x, wm_num);
+        const dim3 thread(WARP_SIZE, THREAD_PER_GRID / WARP_SIZE);
+        const IdxType size_div_w = size / WORD_SIZE;
+        const IdxType size_div_warp = size / WARP_SIZE;
+        assert(size % WARP_SIZE == 0);
+        constexpr int ThreadsDimY = THREAD_PER_GRID / WARP_SIZE;
+
+
+        const int nsum_scan_buf_len = get_nsum_scan_buf_len(block_pair_num); // same grid_x
+
+
+#define CALC_SRCB_SIZE(SrcT)   (sizeof(SrcT) * (ThreadsDimY * (WARP_SIZE - 1) * WARP_SIZE))
+        constexpr int SRCB_S_T = CALC_SRCB_SIZE(T);
+        constexpr int SRCB_S_8 = CALC_SRCB_SIZE(uint8_t);
+#undef CALC_SRCB_SIZE
+        constexpr int BLOCK_SHMEM_KB = SHMEM_USE_KB * THREAD_PER_GRID / 1024;
+        constexpr int DSTB_S_T = WaveletMatrixMultiCu4G_get_dstbuf_kb<BLOCK_SHMEM_KB, MIN_DSTBUF_KB>(SRCB_S_T);
+        constexpr int DSTB_S_8 = WaveletMatrixMultiCu4G_get_dstbuf_kb<BLOCK_SHMEM_KB, MIN_DSTBUF_KB>(SRCB_S_8);
+        static_assert(SHMEM_USE_KB >= 64 || THREAD_PER_GRID == 1024, "if SHMEM_USE_KB < 64, THREAD_PER_GRID must 1024");
+        static_assert(SRCB_S_T + DSTB_S_T <= BLOCK_SHMEM_KB * 1024 && ((DSTB_S_T == 0 && SRCB_S_T + MIN_DSTBUF_KB * 1024> BLOCK_SHMEM_KB * 1024) || (DSTB_S_T >= MIN_DSTBUF_KB * 1024 && SRCB_S_T + DSTB_S_T * 2 > BLOCK_SHMEM_KB * 1024)), "");
+        static_assert(SRCB_S_8 + DSTB_S_8 <= BLOCK_SHMEM_KB * 1024 && ((DSTB_S_8 == 0 && SRCB_S_8 + MIN_DSTBUF_KB * 1024> BLOCK_SHMEM_KB * 1024) || (DSTB_S_8 >= MIN_DSTBUF_KB * 1024 && SRCB_S_8 + DSTB_S_8 * 2 > BLOCK_SHMEM_KB * 1024)), "");
+
+        {   using SrcT = T; using DstT = T;
+            cudaFuncSetAttribute(&WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_T, DSTB_S_T, SrcT, DstT, BlockT, IdxType>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_T + DSTB_S_T);
+        } { using SrcT = T; using DstT = uint8_t;
+            cudaFuncSetAttribute(&WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_T, DSTB_S_T, SrcT, DstT, BlockT, IdxType>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_T + DSTB_S_T);
+        } { using SrcT = uint8_t; using DstT = uint8_t;
+            cudaFuncSetAttribute(&WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_8, DSTB_S_8, SrcT, DstT, BlockT, IdxType>, cudaFuncAttributeMaxDynamicSharedMemorySize, SRCB_S_8 + DSTB_S_8);
+        }
+
+        T* now_cu = src_cu;
+        T* nxt_cu = src_cu + size;
+        IdxType* nsum_zeros_buf = (IdxType*)(nxt_cu + size);
+        IdxType* nsum_zeros_buf2 = nsum_zeros_buf + nsum_scan_buf_len;
+
+        const uint32_t nsum_pos = get_nsum_pos();
+
+        int h = bit_len - 1;
+        if (run_first) {
+            WaveletMatrixMultiCu4G_first_gpu<ThreadsDimY> <<<grid, thread, 0, main_stream >>> (T(mask / 2), block_pair_num, size_div_warp, src_cu, nsum_zeros_buf, buf_byte_div32);
+        }
+        WaveletMatrixMultiCu4G_ExclusiveSum<MAX_BLOCK_X, IdxType> <<< wm_num, grid_x, 0, main_stream >>> (nsum_zeros_buf, nsum_zeros_buf2, get_bv_block_cu(h) + nsum_pos, buf_byte_div32, bv_block_byte_div32);
+
+        for (; h > 8; --h) {
+            using SrcT = T;
+            using DstT = T;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_T, DSTB_S_T> <<<grid, thread, SRCB_S_T + DSTB_S_T, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_zeros_buf, nsum_zeros_buf2, bv_block_byte_div32, buf_byte_div32);
+
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrixMultiCu4G_ExclusiveSum<MAX_BLOCK_X, IdxType> <<< wm_num, grid_x, 0, main_stream >>> (nsum_zeros_buf2, nsum_zeros_buf, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_zeros_buf, nsum_zeros_buf2);
+            swap(now_cu, nxt_cu);
+        }
+        if (h == 8 || (is_same<T, uint16_t>::value && bit_len <= 8 && h == bit_len - 1)) {
+            using SrcT = T;
+            using DstT = uint8_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_T, DSTB_S_T> <<<grid, thread, SRCB_S_T + DSTB_S_T, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_zeros_buf, nsum_zeros_buf2, bv_block_byte_div32, buf_byte_div32);
+            if (h == 0) return;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrixMultiCu4G_ExclusiveSum<MAX_BLOCK_X, IdxType> <<< wm_num, grid_x, 0, main_stream >>> (nsum_zeros_buf2, nsum_zeros_buf, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_zeros_buf, nsum_zeros_buf2);
+            swap(now_cu, nxt_cu);
+            --h;
+        }
+
+        for (; h >= 0; --h) {
+            using SrcT = uint8_t;
+            using DstT = uint8_t;
+            mask >>= 1;
+            BlockT* bv_block_nbit_cu_h = get_bv_block_cu(h);
+            WaveletMatrixMultiCu4G_UpSweep_gpu<ThreadsDimY, SRCB_S_8, DSTB_S_8> <<<grid, thread, SRCB_S_8 + DSTB_S_8, main_stream >>> ((SrcT)mask, block_pair_num, size_div_w, (SrcT*)now_cu, (DstT*)nxt_cu, bv_block_nbit_cu_h, nsum_zeros_buf, nsum_zeros_buf2, bv_block_byte_div32, buf_byte_div32);
+            if (h == 0) break;
+            BlockT* nsum_p = get_bv_block_cu(h - 1) + nsum_pos;
+            WaveletMatrixMultiCu4G_ExclusiveSum<MAX_BLOCK_X, IdxType> <<< wm_num, grid_x, 0, main_stream >>> (nsum_zeros_buf2, nsum_zeros_buf, nsum_p, buf_byte_div32, bv_block_byte_div32);
+            swap(nsum_zeros_buf, nsum_zeros_buf2);
+            swap(now_cu, nxt_cu);
+        }
+    }
+
+    IdxType get_nsum_pos() const {
+        const IdxType size_div_w = size / WORD_SIZE;
+        return size_div_w;
+    }
+    IdxType get_bv_block_h_byte_div32() const {
+        return (bv_block_len * (sizeof(WordT) + sizeof(IdxType))) / 32u;
+    }
+};
+
+} // end namespace wavelet_median
+}}} //end namespace cv::cuda::device
+
+#endif
+#endif // __OPENCV_WAVELET_MATRIX_MULTI_CUH__
diff --git a/modules/cudafilters/src/filtering.cpp b/modules/cudafilters/src/filtering.cpp
index daab3acde10..2ae789c856d 100644
--- a/modules/cudafilters/src/filtering.cpp
+++ b/modules/cudafilters/src/filtering.cpp
@@ -72,6 +72,8 @@ Ptr<Filter> cv::cuda::createColumnSumFilter(int, int, int, int, int, Scalar) { t
 Ptr<Filter> cv::cuda::createMedianFilter(int srcType, int _windowSize, int _partitions){ throw_no_cuda(); return Ptr<Filter>();}
 
 #else
+#include <cuda.h>
+#include <cuda_runtime_api.h>
 
 namespace
 {
@@ -1047,12 +1049,20 @@ Ptr<Filter> cv::cuda::createColumnSumFilter(int srcType, int dstType, int ksize,
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 // Median Filter
 
+// The CUB library is used for the Median Filter with Wavelet Matrix,
+// which has become a standard library since CUDA 11.
+#include "cuda/wavelet_matrix_feature_support_checks.h"
 
 
 namespace cv { namespace cuda { namespace device
 {
     void medianFiltering_gpu(const PtrStepSzb src, PtrStepSzb dst, PtrStepSzi devHist,
         PtrStepSzi devCoarseHist,int kernel, int partitions, cudaStream_t stream);
+
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+    template<typename T>
+    void medianFiltering_wavelet_matrix_gpu(const PtrStepSz<T> src, PtrStepSz<T> dst, int radius, const int num_channels, cudaStream_t stream);
+#endif
 }}}
 
 namespace
@@ -1074,7 +1084,15 @@ namespace
     MedianFilter::MedianFilter(int srcType, int _windowSize, int _partitions) :
         windowSize(_windowSize),partitions(_partitions)
     {
-        CV_Assert( srcType == CV_8UC1 );
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+        CV_Assert(srcType == CV_8UC1  || srcType == CV_8UC3  || srcType == CV_8UC4
+               || srcType == CV_16UC1 || srcType == CV_16UC3 || srcType == CV_16UC4
+               || srcType == CV_32FC1 || srcType == CV_32FC3 || srcType == CV_32FC4);
+#else
+        if (srcType != CV_8UC1) {
+            CV_Error(Error::StsNotImplemented, "If CUDA version is below 10, only implementations that support CV_8UC1 are available");
+        }
+#endif
         CV_Assert(windowSize>=3);
         CV_Assert(_partitions>=1);
 
@@ -1094,6 +1112,18 @@ namespace
         // Kernel needs to be half window size
         int kernel=windowSize/2;
 
+#ifdef __OPENCV_USE_WAVELET_MATRIX_FOR_MEDIAN_FILTER_CUDA__
+        const int depth = src.depth();
+        if (depth == CV_8U) {
+            medianFiltering_wavelet_matrix_gpu<uint8_t>(src, dst, kernel, src.channels(), StreamAccessor::getStream(_stream));
+        } else if (depth == CV_16U) {
+            medianFiltering_wavelet_matrix_gpu<uint16_t>(src, dst, kernel, src.channels(), StreamAccessor::getStream(_stream));
+        } else if (depth == CV_32F) {
+            medianFiltering_wavelet_matrix_gpu<float>(src, dst, kernel, src.channels(), StreamAccessor::getStream(_stream));
+        } else {
+            CV_Assert(depth == CV_8U || depth == CV_16U || depth == CV_32F);
+        }
+#else
         CV_Assert(kernel < src.rows);
         CV_Assert(kernel < src.cols);
 
@@ -1107,6 +1137,7 @@ namespace
         devCoarseHist.setTo(0, _stream);
 
         medianFiltering_gpu(src,dst,devHist, devCoarseHist,kernel,partitions,StreamAccessor::getStream(_stream));
+# endif
     }
 }
 
diff --git a/modules/cudafilters/test/test_filters.cpp b/modules/cudafilters/test/test_filters.cpp
index bb235dad093..432b5d2a5ac 100644
--- a/modules/cudafilters/test/test_filters.cpp
+++ b/modules/cudafilters/test/test_filters.cpp
@@ -43,6 +43,7 @@
 #include "test_precomp.hpp"
 
 #ifdef HAVE_CUDA
+#include "../src/cuda/wavelet_matrix_feature_support_checks.h"
 
 namespace opencv_test { namespace {
 
@@ -647,7 +648,7 @@ INSTANTIATE_TEST_CASE_P(CUDA_Filters, MorphEx, testing::Combine(
 // Median
 
 
-PARAM_TEST_CASE(Median, cv::cuda::DeviceInfo, cv::Size, MatDepth,  KernelSize, UseRoi)
+PARAM_TEST_CASE(Median, cv::cuda::DeviceInfo, cv::Size, MatType, KernelSize, UseRoi)
 {
     cv::cuda::DeviceInfo devInfo;
     cv::Size size;
@@ -681,7 +682,7 @@ CUDA_TEST_P(Median, Accuracy)
     cv::Mat dst_gold;
     cv::medianBlur(src,dst_gold,kernel);
 
-    cv::Rect rect(kernel+1,0,src.cols-(2*kernel+1),src.rows);
+    cv::Rect rect(kernel/2, kernel/2, src.cols-(kernel-1), src.rows-(kernel-1));
     cv::Mat dst_gold_no_border = dst_gold(rect);
     cv::cuda::GpuMat dst_no_border = cv::cuda::GpuMat(dst, rect);
 
@@ -703,6 +704,17 @@ INSTANTIATE_TEST_CASE_P(CUDA_Filters, Median, testing::Combine(
     WHOLE_SUBMAT)
     );
 
-}} // namespace
+INSTANTIATE_TEST_CASE_P(CUDA_Filters_Median_HDR, Median, testing::Combine(
+    ALL_DEVICES,
+    DIFFERENT_SIZES,
+    testing::Values(
+        MatType(CV_8UC3), MatType(CV_8UC4),
+        MatType(CV_16U), MatType(CV_16UC3), MatType(CV_16UC4),
+        MatType(CV_32F), MatType(CV_32FC3), MatType(CV_32FC4)),
+    testing::Values(KernelSize(3), KernelSize(5)),
+    WHOLE_SUBMAT)
+    );
+
 
+}} // namespace
 #endif // HAVE_CUDA

From 8c16a489a1fa9e2ccf21ce8eeb7fbb1affe548b1 Mon Sep 17 00:00:00 2001
From: Vincent Rabaud <vrabaud@google.com>
Date: Thu, 23 May 2024 22:14:29 +0200
Subject: [PATCH 09/12] Get code to compile with CUDA 12.4

This fixes https://github.com/opencv/opencv_contrib/issues/3741
---
 modules/cudaimgproc/src/histogram.cpp | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

diff --git a/modules/cudaimgproc/src/histogram.cpp b/modules/cudaimgproc/src/histogram.cpp
index 177bf75b1ac..51a5ce1a83e 100644
--- a/modules/cudaimgproc/src/histogram.cpp
+++ b/modules/cudaimgproc/src/histogram.cpp
@@ -281,8 +281,13 @@ cv::Ptr<cv::cuda::CLAHE> cv::cuda::createCLAHE(double clipLimit, cv::Size tileGr
 
 namespace
 {
+#if (CUDA_VERSION >= 12040)
+    typedef NppStatus (*get_buf_size_c1_t)(NppiSize oSizeROI, int nLevels, size_t* hpBufferSize);
+    typedef NppStatus (*get_buf_size_c4_t)(NppiSize oSizeROI, int nLevels[], size_t* hpBufferSize);
+#else
     typedef NppStatus (*get_buf_size_c1_t)(NppiSize oSizeROI, int nLevels, int* hpBufferSize);
     typedef NppStatus (*get_buf_size_c4_t)(NppiSize oSizeROI, int nLevels[], int* hpBufferSize);
+#endif
 
     template<int SDEPTH> struct NppHistogramEvenFuncC1
     {
@@ -315,7 +320,11 @@ namespace
             sz.width = src.cols;
             sz.height = src.rows;
 
+#if (CUDA_VERSION >= 12040)
+            size_t buf_size;
+#else
             int buf_size;
+#endif
             get_buf_size(sz, levels, &buf_size);
 
             BufferPool pool(stream);
@@ -349,7 +358,11 @@ namespace
 
             Npp32s* pHist[] = {hist[0].ptr<Npp32s>(), hist[1].ptr<Npp32s>(), hist[2].ptr<Npp32s>(), hist[3].ptr<Npp32s>()};
 
+#if (CUDA_VERSION >= 12040)
+            size_t buf_size;
+#else
             int buf_size;
+#endif
             get_buf_size(sz, levels, &buf_size);
 
             BufferPool pool(stream);
@@ -419,7 +432,11 @@ namespace
             sz.width = src.cols;
             sz.height = src.rows;
 
+#if (CUDA_VERSION >= 12040)
+            size_t buf_size;
+#else
             int buf_size;
+#endif
             get_buf_size(sz, levels.cols, &buf_size);
 
             BufferPool pool(stream);
@@ -460,7 +477,11 @@ namespace
             sz.width = src.cols;
             sz.height = src.rows;
 
+#if (CUDA_VERSION >= 12040)
+            size_t buf_size;
+#else
             int buf_size;
+#endif
             get_buf_size(sz, nLevels, &buf_size);
 
             BufferPool pool(stream);

From d131e7a991e05a02e1724b29b5ed82c346b75f59 Mon Sep 17 00:00:00 2001
From: Vincent Rabaud <vrabaud@google.com>
Date: Thu, 30 May 2024 12:41:21 +0200
Subject: [PATCH 10/12] Merge pull request #3743 from vrabaud:thinning

Misc thinning fixes. #3743

- edges could be modified
- 2 sets of iterations were always done even if the first set did not return any change
- countNonZero is slow compared to hasNonZero

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [x] There is a reference to the original bug report and related work
- [x] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
---
 modules/ximgproc/src/thinning.cpp       | 11 +++++++++--
 modules/ximgproc/test/test_thinning.cpp | 17 +++++++++++------
 2 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/modules/ximgproc/src/thinning.cpp b/modules/ximgproc/src/thinning.cpp
index 00017fe0acb..76e8ad15425 100644
--- a/modules/ximgproc/src/thinning.cpp
+++ b/modules/ximgproc/src/thinning.cpp
@@ -96,6 +96,10 @@ static void thinningIteration(Mat img, int iter, int thinningType){
     Mat marker = Mat::zeros(img.size(), CV_8UC1);
     int rows = img.rows;
     int cols = img.cols;
+    marker.col(0).setTo(1);
+    marker.col(cols - 1).setTo(1);
+    marker.row(0).setTo(1);
+    marker.row(rows - 1).setTo(1);
 
     if(thinningType == THINNING_ZHANGSUEN){
         marker.forEach<uchar>([=](uchar& value, const int postion[]) {
@@ -133,6 +137,7 @@ static void thinningIteration(Mat img, int iter, int thinningType){
             //int m1 = iter == 0 ? (p2 * p4 * p6) : (p2 * p4 * p8);
             //int m2 = iter == 0 ? (p4 * p6 * p8) : (p2 * p6 * p8);
             //if (A == 1 && (B >= 2 && B <= 6) && m1 == 0 && m2 == 0) value = 0;
+            // else value = 1;
         });
     }
     if(thinningType == THINNING_GUOHALL){
@@ -170,6 +175,7 @@ static void thinningIteration(Mat img, int iter, int thinningType){
             //int N  = N1 < N2 ? N1 : N2;
             //int m  = iter == 0 ? ((p6 | p7 | (!p9)) & p8) : ((p2 | p3 | (!p5)) & p4);
             //if ((C == 1) && ((N >= 2) && ((N <= 3)) & (m == 0))) value = 0;
+            // else value = 1;
         });
     }
 
@@ -183,16 +189,17 @@ void thinning(InputArray input, OutputArray output, int thinningType){
     // Enforce the range of the input image to be in between 0 - 255
     processed /= 255;
 
-    Mat prev = Mat::zeros(processed.size(), CV_8UC1);
+    Mat prev = processed.clone();
     Mat diff;
 
     do {
         thinningIteration(processed, 0, thinningType);
         thinningIteration(processed, 1, thinningType);
         absdiff(processed, prev, diff);
+        if (!hasNonZero(diff)) break;
         processed.copyTo(prev);
     }
-    while (countNonZero(diff) > 0);
+    while (true);
 
     processed *= 255;
 
diff --git a/modules/ximgproc/test/test_thinning.cpp b/modules/ximgproc/test/test_thinning.cpp
index 733fe85d473..7d5c5ac480f 100644
--- a/modules/ximgproc/test/test_thinning.cpp
+++ b/modules/ximgproc/test/test_thinning.cpp
@@ -6,9 +6,12 @@
 
 namespace opencv_test { namespace {
 
-static int createTestImage(Mat& src)
+static int createTestImage(Mat1b& src)
 {
-    src = Mat::zeros(Size(256, 256), CV_8UC1);
+    src = Mat1b::zeros(Size(256, 256));
+    // Create a corner point that should not be affected.
+    src(0, 0) = 255;
+
     for (int x = 50; x < src.cols - 50; x += 50)
     {
         cv::circle(src, Point(x, x/2), 30 + x/2, Scalar(255), 5);
@@ -20,13 +23,14 @@ static int createTestImage(Mat& src)
 
 TEST(ximgproc_Thinning, simple_ZHANGSUEN)
 {
-    Mat src;
+    Mat1b src;
     int src_pixels = createTestImage(src);
 
-    Mat dst;
+    Mat1b dst;
     thinning(src, dst, THINNING_ZHANGSUEN);
     int dst_pixels = countNonZero(dst);
     EXPECT_LE(dst_pixels, src_pixels);
+    EXPECT_EQ(dst(0, 0), 255);
 
 #if 0
     imshow("src", src); imshow("dst", dst); waitKey();
@@ -35,13 +39,14 @@ TEST(ximgproc_Thinning, simple_ZHANGSUEN)
 
 TEST(ximgproc_Thinning, simple_GUOHALL)
 {
-    Mat src;
+    Mat1b src;
     int src_pixels = createTestImage(src);
 
-    Mat dst;
+    Mat1b dst;
     thinning(src, dst, THINNING_GUOHALL);
     int dst_pixels = countNonZero(dst);
     EXPECT_LE(dst_pixels, src_pixels);
+    EXPECT_EQ(dst(0, 0), 255);
 
 #if 0
     imshow("src", src); imshow("dst", dst); waitKey();

From 1ed3dd2c53888e3289afdb22ec4e9ebbff3dba87 Mon Sep 17 00:00:00 2001
From: Alexander Smorkalov <2536374+asmorkalov@users.noreply.github.com>
Date: Thu, 30 May 2024 15:48:48 +0300
Subject: [PATCH 11/12] Merge pull request #3744 from
 asmorkalov:as/variadic_tuple

Added CUDA 12.4+ support #3744

Tries to fix https://github.com/opencv/opencv_contrib/issues/3690 for CUDA 12.4+
Related patch to main repo: https://github.com/opencv/opencv/pull/25658

Changes:
- Added branches to support new variadic implementation of thrust::tuple
- Added branch with std::array instead of std::tuple in split-merge and grid operations. The new branch got rid of namespace clash: cv::cuda in OpenCV and ::cuda in CUDA standard library (injected by Thrust). Old tuple branches presumed for compatibility with old code and CUDA versions before 12.4.

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [x] There is a reference to the original bug report and related work
- [ ] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [ ] The feature is well documented and sample code can be built with the project CMake
---
 modules/cudaarithm/src/cuda/polar_cart.cu     |  18 +--
 modules/cudaarithm/src/cuda/split_merge.cu    |   9 +-
 .../opencv2/cudev/block/detail/reduce.hpp     |  55 ++++++--
 .../cudev/block/detail/reduce_key_val.hpp     |  62 +++++++++
 .../include/opencv2/cudev/block/reduce.hpp    |  35 +++++
 .../opencv2/cudev/grid/detail/split_merge.hpp |  25 +++-
 .../opencv2/cudev/grid/detail/transform.hpp   | 120 ++++++++++++++++
 .../opencv2/cudev/grid/split_merge.hpp        | 130 +++++++++++++++---
 .../include/opencv2/cudev/grid/transform.hpp  |  91 ++++++++++++
 .../include/opencv2/cudev/ptr2d/glob.hpp      |  12 ++
 .../cudev/include/opencv2/cudev/ptr2d/zip.hpp |  22 +++
 modules/cudev/test/test_split_merge.cu        |   7 +-
 12 files changed, 537 insertions(+), 49 deletions(-)

diff --git a/modules/cudaarithm/src/cuda/polar_cart.cu b/modules/cudaarithm/src/cuda/polar_cart.cu
index 2fb1315e619..12980e424ff 100644
--- a/modules/cudaarithm/src/cuda/polar_cart.cu
+++ b/modules/cudaarithm/src/cuda/polar_cart.cu
@@ -133,23 +133,9 @@ void cv::cuda::cartToPolar(InputArray _x, InputArray _y, OutputArray _mag, Outpu
     GpuMat_<float> anglec(angle.reshape(1));
 
     if (angleInDegrees)
-    {
-        gridTransformTuple(zipPtr(xc, yc),
-                           tie(magc, anglec),
-                           make_tuple(
-                               binaryTupleAdapter<0, 1>(magnitude_func<float>()),
-                               binaryTupleAdapter<0, 1>(direction_func<float, true>())),
-                           stream);
-    }
+        gridTransformBinary(xc, yc, magc, anglec, magnitude_func<float>(), direction_func<float, true>(), stream);
     else
-    {
-        gridTransformTuple(zipPtr(xc, yc),
-                           tie(magc, anglec),
-                           make_tuple(
-                               binaryTupleAdapter<0, 1>(magnitude_func<float>()),
-                               binaryTupleAdapter<0, 1>(direction_func<float, false>())),
-                           stream);
-    }
+        gridTransformBinary(xc, yc, magc, anglec, magnitude_func<float>(), direction_func<float, false>(), stream);
 
     syncOutput(mag, _mag, stream);
     syncOutput(angle, _angle, stream);
diff --git a/modules/cudaarithm/src/cuda/split_merge.cu b/modules/cudaarithm/src/cuda/split_merge.cu
index 5b3af10775d..f0acb840a9e 100644
--- a/modules/cudaarithm/src/cuda/split_merge.cu
+++ b/modules/cudaarithm/src/cuda/split_merge.cu
@@ -67,7 +67,8 @@ namespace
     {
         static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
         {
-            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1])),
+            const std::array<GlobPtrSz<T>, 2> d_src = {globPtr<T>(src[0]), globPtr<T>(src[1])};
+            gridMerge(d_src,
                     globPtr<typename MakeVec<T, 2>::type>(dst),
                     stream);
         }
@@ -77,7 +78,8 @@ namespace
     {
         static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
         {
-            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2])),
+            const std::array<GlobPtrSz<T>, 3> d_src = {globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2])};
+            gridMerge(d_src,
                     globPtr<typename MakeVec<T, 3>::type>(dst),
                     stream);
         }
@@ -87,7 +89,8 @@ namespace
     {
         static void call(const GpuMat* src, GpuMat& dst, Stream& stream)
         {
-            gridMerge(zipPtr(globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2]), globPtr<T>(src[3])),
+            const std::array<GlobPtrSz<T>, 4 > d_src = {globPtr<T>(src[0]), globPtr<T>(src[1]), globPtr<T>(src[2]), globPtr<T>(src[3])};
+            gridMerge(d_src,
                     globPtr<typename MakeVec<T, 4>::type>(dst),
                     stream);
         }
diff --git a/modules/cudev/include/opencv2/cudev/block/detail/reduce.hpp b/modules/cudev/include/opencv2/cudev/block/detail/reduce.hpp
index 151e949a617..5bd1737aa3c 100644
--- a/modules/cudev/include/opencv2/cudev/block/detail/reduce.hpp
+++ b/modules/cudev/include/opencv2/cudev/block/detail/reduce.hpp
@@ -154,6 +154,17 @@ namespace block_reduce_detail
         val = smem[tid];
     }
 
+
+    // merge
+
+    template <typename T, class Op>
+    __device__ __forceinline__ void merge(volatile T* smem, T& val, uint tid, uint delta, const Op& op)
+    {
+        T reg = smem[tid + delta];
+        smem[tid] = val = op(val, reg);
+    }
+
+#if (CUDART_VERSION < 12040)
     template <typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
               typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9>
     __device__ __forceinline__ void loadToSmem(const tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem,
@@ -172,15 +183,6 @@ namespace block_reduce_detail
         For<0, tuple_size<tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9> >::value>::loadFromSmem(smem, val, tid);
     }
 
-    // merge
-
-    template <typename T, class Op>
-    __device__ __forceinline__ void merge(volatile T* smem, T& val, uint tid, uint delta, const Op& op)
-    {
-        T reg = smem[tid + delta];
-        smem[tid] = val = op(val, reg);
-    }
-
     template <typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
               typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9,
               class Op0, class Op1, class Op2, class Op3, class Op4, class Op5, class Op6, class Op7, class Op8, class Op9>
@@ -214,6 +216,41 @@ namespace block_reduce_detail
     }
 #endif
 
+#else
+    template <typename... P, typename... R>
+    __device__ __forceinline__ void loadToSmem(const tuple<P...>& smem, const tuple<R...>& val, uint tid)
+    {
+        For<0, tuple_size<tuple<P...> >::value>::loadToSmem(smem, val, tid);
+    }
+
+    template <typename... P, typename... R>
+    __device__ __forceinline__ void loadFromSmem(const tuple<P...>& smem, const tuple<R...>& val, uint tid)
+    {
+        For<0, tuple_size<tuple<P...> >::value>::loadFromSmem(smem, val, tid);
+    }
+
+    template <typename... P, typename... R, class... Op>
+    __device__ __forceinline__ void merge(const tuple<P...>& smem, const tuple<R...>& val, uint tid, uint delta, const tuple<Op...>& op)
+    {
+        For<0, tuple_size<tuple<P...> >::value>::merge(smem, val, tid, delta, op);
+    }
+
+    // mergeShfl
+
+    template <typename T, class Op>
+    __device__ __forceinline__ void mergeShfl(T& val, uint delta, uint width, const Op& op)
+    {
+        T reg = shfl_down(val, delta, width);
+        val = op(val, reg);
+    }
+
+    template <typename... R, class... Op>
+    __device__ __forceinline__ void mergeShfl(const tuple<R...>& val, uint delta, uint width, const tuple<Op...>& op)
+    {
+        For<0, tuple_size<tuple<R...> >::value>::mergeShfl(val, delta, width, op);
+    }
+#endif
+
     // Generic
 
     template <int N> struct Generic
diff --git a/modules/cudev/include/opencv2/cudev/block/detail/reduce_key_val.hpp b/modules/cudev/include/opencv2/cudev/block/detail/reduce_key_val.hpp
index 4af834a446e..43876decc92 100644
--- a/modules/cudev/include/opencv2/cudev/block/detail/reduce_key_val.hpp
+++ b/modules/cudev/include/opencv2/cudev/block/detail/reduce_key_val.hpp
@@ -160,6 +160,7 @@ namespace block_reduce_key_val_detail
         data = smem[tid];
     }
 
+#if (CUDART_VERSION < 12040)
     template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
               typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
     __device__ __forceinline__ void loadToSmem(const tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& smem,
@@ -241,6 +242,67 @@ namespace block_reduce_key_val_detail
     {
         For<0, tuple_size<tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::merge(skeys, key, svals, val, cmp, tid, delta);
     }
+#else
+    template <typename... VP, typename... VR>
+    __device__ __forceinline__ void loadToSmem(const tuple<VP...>& smem, const tuple<VR...>& data, uint tid)
+    {
+        For<0, tuple_size<tuple<VP...> >::value>::loadToSmem(smem, data, tid);
+    }
+
+    template <typename... VP, typename... VR>
+    __device__ __forceinline__ void loadFromSmem(const tuple<VP...>& smem, const tuple<VR...>& data, uint tid)
+    {
+        For<0, tuple_size<tuple<VP...> >::value>::loadFromSmem(smem, data, tid);
+    }
+
+    // copyVals
+
+    template <typename V>
+    __device__ __forceinline__ void copyVals(volatile V* svals, V& val, uint tid, uint delta)
+    {
+        svals[tid] = val = svals[tid + delta];
+    }
+
+    template <typename... VP, typename... VR>
+    __device__ __forceinline__ void copyVals(const tuple<VP...>& svals, const tuple<VR...>& val, uint tid, uint delta)
+    {
+        For<0, tuple_size<tuple<VP...> >::value>::copy(svals, val, tid, delta);
+    }
+
+    // merge
+
+    template <typename K, typename V, class Cmp>
+    __device__ void merge(volatile K* skeys, K& key, volatile V* svals, V& val, const Cmp& cmp, uint tid, uint delta)
+    {
+        K reg = skeys[tid + delta];
+
+        if (cmp(reg, key))
+        {
+            skeys[tid] = key = reg;
+            copyVals(svals, val, tid, delta);
+        }
+    }
+
+    template <typename K, typename... VP, typename... VR, class Cmp>
+    __device__ void merge(volatile K* skeys, K& key, const tuple<VP...>& svals, const tuple<VR...>& val, const Cmp& cmp, uint tid, uint delta)
+    {
+        K reg = skeys[tid + delta];
+
+        if (cmp(reg, key))
+        {
+            skeys[tid] = key = reg;
+            copyVals(svals, val, tid, delta);
+        }
+    }
+
+    template <typename... KP, typename... KR, typename... VP, typename... VR, class... Cmp>
+    __device__ __forceinline__ void merge(const tuple<KP...>& skeys, const tuple<KR...>& key,
+                                          const tuple<VP...>& svals, const tuple<VR...>& val,
+                                          const tuple<Cmp...>& cmp, uint tid, uint delta)
+    {
+        For<0, tuple_size<tuple<VP...> >::value>::merge(skeys, key, svals, val, cmp, tid, delta);
+    }
+#endif
 
     // Generic
 
diff --git a/modules/cudev/include/opencv2/cudev/block/reduce.hpp b/modules/cudev/include/opencv2/cudev/block/reduce.hpp
index 06f59a16ae9..9dde278da84 100644
--- a/modules/cudev/include/opencv2/cudev/block/reduce.hpp
+++ b/modules/cudev/include/opencv2/cudev/block/reduce.hpp
@@ -51,6 +51,7 @@
 #include "../warp/reduce.hpp"
 #include "detail/reduce.hpp"
 #include "detail/reduce_key_val.hpp"
+#include <cuda_runtime_api.h>
 
 namespace cv { namespace cudev {
 
@@ -65,6 +66,7 @@ __device__ __forceinline__ void blockReduce(volatile T* smem, T& val, uint tid,
     block_reduce_detail::Dispatcher<N>::reductor::template reduce<volatile T*, T&, const Op&>(smem, val, tid, op);
 }
 
+#if (CUDART_VERSION < 12040)
 template <int N,
           typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
           typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9,
@@ -126,6 +128,39 @@ __device__ __forceinline__ void blockReduceKeyVal(const tuple<KP0, KP1, KP2, KP3
             >(skeys, key, svals, val, tid, cmp);
 }
 
+#else
+
+template <int N, typename... P, typename... R, typename... Op>
+__device__ __forceinline__ void blockReduce(const tuple<P...>& smem,
+                                            const tuple<R...>& val,
+                                            uint tid,
+                                            const tuple<Op...>& op)
+{
+    block_reduce_detail::Dispatcher<N>::reductor::template reduce<const tuple<P...>&, const tuple<R...>&, const tuple<Op...>&>(smem, val, tid, op);
+}
+
+// blockReduceKeyVal
+
+template <int N, typename K, typename V, class Cmp>
+__device__ __forceinline__ void blockReduceKeyVal(volatile K* skeys, K& key, volatile V* svals, V& val, uint tid, const Cmp& cmp)
+{
+    block_reduce_key_val_detail::Dispatcher<N>::reductor::template reduce<volatile K*, K&, volatile V*, V&, const Cmp&>(skeys, key, svals, val, tid, cmp);
+}
+
+template <int N, typename K, typename... VP, typename... VR, class Cmp>
+__device__ __forceinline__ void blockReduceKeyVal(volatile K* skeys, K& key, const tuple<VP...>& svals, const tuple<VR...>& val, uint tid, const Cmp& cmp)
+{
+    block_reduce_key_val_detail::Dispatcher<N>::reductor::template reduce<volatile K*, K&, const tuple<VP...>&, const tuple<VR...>&, const Cmp&>(skeys, key, svals, val, tid, cmp);
+}
+
+template <int N, typename... KP, typename... KR, typename... VP, typename... VR, class... Cmp>
+__device__ __forceinline__ void blockReduceKeyVal(const tuple<KP...>& skeys, const tuple<KR...>& key, const tuple<VP...>& svals, const tuple<VR...>& val, uint tid, const tuple<Cmp...>& cmp)
+{
+    block_reduce_key_val_detail::Dispatcher<N>::reductor::template reduce< const tuple<KP...>&, const tuple<KR...>&, const tuple<VP...>&, const tuple<VR...>&, const tuple<Cmp...>&>(skeys, key, svals, val, tid, cmp);
+}
+
+#endif
+
 //! @}
 
 }}
diff --git a/modules/cudev/include/opencv2/cudev/grid/detail/split_merge.hpp b/modules/cudev/include/opencv2/cudev/grid/detail/split_merge.hpp
index 3f512060165..df8bed3a948 100644
--- a/modules/cudev/include/opencv2/cudev/grid/detail/split_merge.hpp
+++ b/modules/cudev/include/opencv2/cudev/grid/detail/split_merge.hpp
@@ -157,28 +157,47 @@ namespace grid_split_merge_detail
     template <class Policy> struct MergeImpl<2, Policy>
     {
         template <class SrcPtrTuple, typename DstType, class MaskPtr>
-        __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        __host__ static void mergeTuple(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
         {
             mergeC2<Policy>(get<0>(src), get<1>(src), dst, mask, rows, cols, stream);
         }
+
+        template <class SrcPtrArray, typename DstType, class MaskPtr>
+        __host__ static void mergeArray(const SrcPtrArray& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        {
+            mergeC2<Policy>(src[0], src[1], dst, mask, rows, cols, stream);
+        }
+
     };
 
     template <class Policy> struct MergeImpl<3, Policy>
     {
         template <class SrcPtrTuple, typename DstType, class MaskPtr>
-        __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        __host__ static void mergeTuple(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
         {
             mergeC3<Policy>(get<0>(src), get<1>(src), get<2>(src), dst, mask, rows, cols, stream);
         }
+
+        template <class SrcPtrArray, typename DstType, class MaskPtr>
+        __host__ static void mergeArray(const SrcPtrArray& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        {
+            mergeC3<Policy>(src[0], src[1], src[2], dst, mask, rows, cols, stream);
+        }
     };
 
     template <class Policy> struct MergeImpl<4, Policy>
     {
         template <class SrcPtrTuple, typename DstType, class MaskPtr>
-        __host__ static void merge(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        __host__ static void mergeTuple(const SrcPtrTuple& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
         {
             mergeC4<Policy>(get<0>(src), get<1>(src), get<2>(src), get<3>(src), dst, mask, rows, cols, stream);
         }
+
+        template <class SrcPtrArray, typename DstType, class MaskPtr>
+        __host__ static void mergeArray(const SrcPtrArray& src, const GlobPtr<DstType>& dst, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        {
+            mergeC4<Policy>(src[0], src[1], src[2], src[3], dst, mask, rows, cols, stream);
+        }
     };
 
     // split
diff --git a/modules/cudev/include/opencv2/cudev/grid/detail/transform.hpp b/modules/cudev/include/opencv2/cudev/grid/detail/transform.hpp
index 557797d7c85..4e901ac751d 100644
--- a/modules/cudev/include/opencv2/cudev/grid/detail/transform.hpp
+++ b/modules/cudev/include/opencv2/cudev/grid/detail/transform.hpp
@@ -179,6 +179,23 @@ namespace grid_transform_detail
         dst(y, x) = saturate_cast<DstType>(op(src1(y, x), src2(y, x)));
     }
 
+    // transformSimple, 2 outputs
+    // The overloads are added for polar_cart.cu to compute magnitude and phase with single call
+    // the previous implementation with touple causes cuda namespace clash. See https://github.com/opencv/opencv_contrib/issues/3690
+    template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+    __global__ void transformSimple(const SrcPtr1 src1, const SrcPtr2 src2, GlobPtr<DstType1> dst1, GlobPtr<DstType2> dst2,
+                                    const BinOp1 op1, const BinOp2 op2, const MaskPtr mask, const int rows, const int cols)
+    {
+        const int x = blockIdx.x * blockDim.x + threadIdx.x;
+        const int y = blockIdx.y * blockDim.y + threadIdx.y;
+
+        if (x >= cols || y >= rows || !mask(y, x))
+            return;
+
+        dst1(y, x) = saturate_cast<DstType1>(op1(src1(y, x), src2(y, x)));
+        dst2(y, x) = saturate_cast<DstType2>(op2(src1(y, x), src2(y, x)));
+    }
+
     // transformSmart
 
     template <int SHIFT, typename SrcType, typename DstType, class UnOp, class MaskPtr>
@@ -248,6 +265,52 @@ namespace grid_transform_detail
         }
     }
 
+    // transformSmart, 2 outputs
+    // The overloads are added for polar_cart.cu to compute magnitude and phase with single call
+    // the previous implementation with touple causes cuda namespace clash. See https://github.com/opencv/opencv_contrib/issues/3690
+    template <int SHIFT, typename SrcType1, typename SrcType2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+    __global__ void transformSmart(const GlobPtr<SrcType1> src1_, const GlobPtr<SrcType2> src2_,
+                                   GlobPtr<DstType1> dst1_, GlobPtr<DstType2> dst2_,
+                                   const BinOp1 op1, const BinOp2 op2, const MaskPtr mask, const int rows, const int cols)
+    {
+        typedef typename MakeVec<SrcType1, SHIFT>::type read_type1;
+        typedef typename MakeVec<SrcType2, SHIFT>::type read_type2;
+        typedef typename MakeVec<DstType1, SHIFT>::type write_type1;
+        typedef typename MakeVec<DstType2, SHIFT>::type write_type2;
+
+        const int x = blockIdx.x * blockDim.x + threadIdx.x;
+        const int y = blockIdx.y * blockDim.y + threadIdx.y;
+        const int x_shifted = x * SHIFT;
+
+        if (y < rows)
+        {
+            const SrcType1* src1 = src1_.row(y);
+            const SrcType2* src2 = src2_.row(y);
+            DstType1* dst1 = dst1_.row(y);
+            DstType2* dst2 = dst2_.row(y);
+
+            if (x_shifted + SHIFT - 1 < cols)
+            {
+                const read_type1 src1_n_el = ((const read_type1*)src1)[x];
+                const read_type2 src2_n_el = ((const read_type2*)src2)[x];
+
+                OpUnroller<SHIFT>::unroll(src1_n_el, src2_n_el, ((write_type1*)dst1)[x], op1, mask, x_shifted, y);
+                OpUnroller<SHIFT>::unroll(src1_n_el, src2_n_el, ((write_type2*)dst2)[x], op2, mask, x_shifted, y);
+            }
+            else
+            {
+                for (int real_x = x_shifted; real_x < cols; ++real_x)
+                {
+                    if (mask(y, real_x))
+                    {
+                        dst1[real_x] = op1(src1[real_x], src2[real_x]);
+                        dst2[real_x] = op2(src1[real_x], src2[real_x]);
+                    }
+                }
+            }
+        }
+    }
+
     // TransformDispatcher
 
     template <bool UseSmart, class Policy> struct TransformDispatcher;
@@ -279,6 +342,20 @@ namespace grid_transform_detail
             if (stream == 0)
                 CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() );
         }
+
+        template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+        __host__ static void call(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtr<DstType1>& dst1, const GlobPtr<DstType2>& dst2,
+                                  const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        {
+            const dim3 block(Policy::block_size_x, Policy::block_size_y);
+            const dim3 grid(divUp(cols, block.x), divUp(rows, block.y));
+
+            transformSimple<<<grid, block, 0, stream>>>(src1, src2, dst1, dst2, op1, op2, mask, rows, cols);
+            CV_CUDEV_SAFE_CALL( cudaGetLastError() );
+
+            if (stream == 0)
+                CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() );
+        }
     };
 
     template <class Policy> struct TransformDispatcher<true, Policy>
@@ -336,6 +413,33 @@ namespace grid_transform_detail
             if (stream == 0)
                 CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() );
         }
+
+        template <typename SrcType1, typename SrcType2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+        __host__ static void call(const GlobPtr<SrcType1>& src1, const GlobPtr<SrcType2>& src2,
+                                  const GlobPtr<DstType1>& dst1, const GlobPtr<DstType2>& dst2,
+                                  const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+        {
+            if (Policy::shift == 1 ||
+                !isAligned(src1.data, Policy::shift * sizeof(SrcType1))  || !isAligned(src1.step, Policy::shift * sizeof(SrcType1)) ||
+                !isAligned(src2.data, Policy::shift * sizeof(SrcType2))  || !isAligned(src2.step, Policy::shift * sizeof(SrcType2)) ||
+                !isAligned(dst1.data,  Policy::shift * sizeof(DstType1)) || !isAligned(dst1.step,  Policy::shift * sizeof(DstType1))||
+                !isAligned(dst2.data,  Policy::shift * sizeof(DstType2)) || !isAligned(dst2.step,  Policy::shift * sizeof(DstType2))
+            )
+            {
+                TransformDispatcher<false, Policy>::call(src1, src2, dst1, dst2, op1, op2, mask, rows, cols, stream);
+                return;
+            }
+
+            const dim3 block(Policy::block_size_x, Policy::block_size_y);
+            const dim3 grid(divUp(cols, block.x * Policy::shift), divUp(rows, block.y));
+
+            transformSmart<Policy::shift><<<grid, block, 0, stream>>>(src1, src2, dst1, dst2, op1, op2, mask, rows, cols);
+            CV_CUDEV_SAFE_CALL( cudaGetLastError() );
+
+            if (stream == 0)
+                CV_CUDEV_SAFE_CALL( cudaDeviceSynchronize() );
+        }
+
     };
 
     template <class Policy, class SrcPtr, typename DstType, class UnOp, class MaskPtr>
@@ -350,6 +454,13 @@ namespace grid_transform_detail
         TransformDispatcher<false, Policy>::call(src1, src2, dst, op, mask, rows, cols, stream);
     }
 
+    template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+    __host__ void transform_binary(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtr<DstType1>& dst1, const GlobPtr<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+    {
+        TransformDispatcher<false, Policy>::call(src1, src2, dst1, dst2, op1, op2, mask, rows, cols, stream);
+    }
+
     template <class Policy, typename SrcType, typename DstType, class UnOp, class MaskPtr>
     __host__ void transform_unary(const GlobPtr<SrcType>& src, const GlobPtr<DstType>& dst, const UnOp& op, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
     {
@@ -362,6 +473,15 @@ namespace grid_transform_detail
         TransformDispatcher<VecTraits<SrcType1>::cn == 1 && VecTraits<SrcType2>::cn == 1 && VecTraits<DstType>::cn == 1 && Policy::shift != 1, Policy>::call(src1, src2, dst, op, mask, rows, cols, stream);
     }
 
+    template <class Policy, typename SrcType1, typename SrcType2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+    __host__ void transform_binary(const GlobPtr<SrcType1>& src1, const GlobPtr<SrcType2>& src2, const GlobPtr<DstType1>& dst1, const GlobPtr<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, int rows, int cols, cudaStream_t stream)
+    {
+        TransformDispatcher<VecTraits<SrcType1>::cn == 1 && VecTraits<SrcType2>::cn == 1 &&
+                            VecTraits<DstType1>::cn == 1 && VecTraits<DstType2>::cn == 1 &&
+                            Policy::shift != 1, Policy>::call(src1, src2, dst1, dst2, op1, op2, mask, rows, cols, stream);
+    }
+
     // transform_tuple
 
     template <int count> struct Unroll
diff --git a/modules/cudev/include/opencv2/cudev/grid/split_merge.hpp b/modules/cudev/include/opencv2/cudev/grid/split_merge.hpp
index 5c92a813ed8..115d8c55e46 100644
--- a/modules/cudev/include/opencv2/cudev/grid/split_merge.hpp
+++ b/modules/cudev/include/opencv2/cudev/grid/split_merge.hpp
@@ -72,11 +72,11 @@ __host__ void gridMerge_(const SrcPtrTuple& src, GpuMat_<DstType>& dst, const Ma
 
     dst.create(rows, cols);
 
-    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::merge(shrinkPtr(src),
-                                                                              shrinkPtr(dst),
-                                                                              shrinkPtr(mask),
-                                                                              rows, cols,
-                                                                              StreamAccessor::getStream(stream));
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeTuple(shrinkPtr(src),
+                                                                                   shrinkPtr(dst),
+                                                                                   shrinkPtr(mask),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
 }
 
 template <class Policy, class SrcPtrTuple, typename DstType, class MaskPtr>
@@ -90,7 +90,7 @@ __host__ void gridMerge_(const SrcPtrTuple& src, const GlobPtrSz<DstType>& dst,
     CV_Assert( getRows(dst) == rows && getCols(dst) == cols );
     CV_Assert( getRows(mask) == rows && getCols(mask) == cols );
 
-    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::merge(shrinkPtr(src),
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeTuple(shrinkPtr(src),
                                                                               shrinkPtr(dst),
                                                                               shrinkPtr(mask),
                                                                               rows, cols,
@@ -107,11 +107,11 @@ __host__ void gridMerge_(const SrcPtrTuple& src, GpuMat_<DstType>& dst, Stream&
 
     dst.create(rows, cols);
 
-    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::merge(shrinkPtr(src),
-                                                                              shrinkPtr(dst),
-                                                                              WithOutMask(),
-                                                                              rows, cols,
-                                                                              StreamAccessor::getStream(stream));
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeTuple(shrinkPtr(src),
+                                                                                   shrinkPtr(dst),
+                                                                                   WithOutMask(),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
 }
 
 template <class Policy, class SrcPtrTuple, typename DstType>
@@ -124,13 +124,87 @@ __host__ void gridMerge_(const SrcPtrTuple& src, const GlobPtrSz<DstType>& dst,
 
     CV_Assert( getRows(dst) == rows && getCols(dst) == cols );
 
-    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::merge(shrinkPtr(src),
-                                                                              shrinkPtr(dst),
-                                                                              WithOutMask(),
-                                                                              rows, cols,
-                                                                              StreamAccessor::getStream(stream));
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeTuple(shrinkPtr(src),
+                                                                                   shrinkPtr(dst),
+                                                                                   WithOutMask(),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
+}
+
+template <class Policy, class ArrayType, size_t ArraySize, typename DstType, class MaskPtr>
+__host__ void gridMergeArray_(const std::array<ArrayType, ArraySize>& src, GpuMat_<DstType>& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    CV_Assert( VecTraits<DstType>::cn == src.size() );
+
+    const int rows = getRows(src);
+    const int cols = getCols(src);
+
+    CV_Assert( getRows(mask) == rows && getCols(mask) == cols );
+
+    dst.create(rows, cols);
+
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeArray(src,
+                                                                                   shrinkPtr(dst),
+                                                                                   shrinkPtr(mask),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
+}
+
+template <class Policy, class ArrayType, size_t ArraySize, typename DstType, class MaskPtr>
+__host__ void gridMergeArray_(const std::array<ArrayType, ArraySize>& src, const GlobPtrSz<DstType>& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    CV_Assert( VecTraits<DstType>::cn == src.size() );
+
+    const int rows = src[0].rows;
+    const int cols = src[0].cols;
+
+    CV_Assert( getRows(dst) == rows && getCols(dst) == cols );
+    CV_Assert( getRows(mask) == rows && getCols(mask) == cols );
+
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeArray(src,
+                                                                                   shrinkPtr(dst),
+                                                                                   shrinkPtr(mask),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
 }
 
+template <class Policy, class ArrayType, size_t ArraySize, typename DstType>
+__host__ void gridMergeArray_(const std::array<ArrayType, ArraySize>& src, GpuMat_<DstType>& dst, Stream& stream = Stream::Null())
+{
+    CV_Assert( VecTraits<DstType>::cn == src.size() );
+
+    const int rows = src[0].rows;
+    const int cols = src[0].cols;
+
+    dst.create(rows, cols);
+
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeArray(src,
+                                                                                   shrinkPtr(dst),
+                                                                                   WithOutMask(),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
+}
+
+template <class Policy, class ArrayType, size_t ArraySize, typename DstType>
+__host__ void gridMergeArray_(const std::array<ArrayType, ArraySize>& src, const GlobPtrSz<DstType>& dst, Stream& stream = Stream::Null())
+{
+    CV_Assert( VecTraits<DstType>::cn == src.size() );
+
+    const int rows = src[0].rows;
+    const int cols = src[0].cols;
+
+    CV_Assert( getRows(dst) == rows && getCols(dst) == cols );
+
+    grid_split_merge_detail::MergeImpl<VecTraits<DstType>::cn, Policy>::mergeArray(src,
+                                                                                   shrinkPtr(dst),
+                                                                                   WithOutMask(),
+                                                                                   rows, cols,
+                                                                                   StreamAccessor::getStream(stream));
+}
+
+
+///////////////////////////////////////////////////////////////
+
 template <class Policy, class SrcPtr, typename DstType, class MaskPtr>
 __host__ void gridSplit_(const SrcPtr& src, const tuple< GpuMat_<DstType>&, GpuMat_<DstType>& >& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
@@ -522,6 +596,30 @@ __host__ void gridMerge(const SrcPtrTuple& src, const GlobPtrSz<DstType>& dst, S
     gridMerge_<DefaultSplitMergePolicy>(src, dst, stream);
 }
 
+template <class ArrayType, size_t ArraySize, typename DstType, class MaskPtr>
+__host__ void gridMergeArray(const std::array<ArrayType, ArraySize>& src, GpuMat_<DstType>& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    gridMergeArray_<DefaultSplitMergePolicy>(src, dst, mask, stream);
+}
+
+template <class ArrayType, size_t ArraySize, typename DstType, class MaskPtr>
+__host__ void gridMerge(const std::array<ArrayType, ArraySize>& src, const GlobPtrSz<DstType>& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    gridMergeArray_<DefaultSplitMergePolicy>(src, dst, mask, stream);
+}
+
+template <class ArrayType, size_t ArraySize, typename DstType>
+__host__ void gridMerge(const std::array<ArrayType, ArraySize>& src, GpuMat_<DstType>& dst, Stream& stream = Stream::Null())
+{
+    gridMergeArray_<DefaultSplitMergePolicy>(src, dst, stream);
+}
+
+template <class ArrayType, size_t ArraySize, typename DstType>
+__host__ void gridMerge(const std::array<ArrayType, ArraySize>& src, const GlobPtrSz<DstType>& dst, Stream& stream = Stream::Null())
+{
+    gridMergeArray_<DefaultSplitMergePolicy>(src, dst, stream);
+}
+
 template <class SrcPtr, typename DstType, class MaskPtr>
 __host__ void gridSplit(const SrcPtr& src, const tuple< GpuMat_<DstType>&, GpuMat_<DstType>& >& dst, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
diff --git a/modules/cudev/include/opencv2/cudev/grid/transform.hpp b/modules/cudev/include/opencv2/cudev/grid/transform.hpp
index 4f7d191e64b..f89cdf5d484 100644
--- a/modules/cudev/include/opencv2/cudev/grid/transform.hpp
+++ b/modules/cudev/include/opencv2/cudev/grid/transform.hpp
@@ -121,6 +121,22 @@ __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, Gpu
     grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst), op, shrinkPtr(mask), rows, cols, StreamAccessor::getStream(stream));
 }
 
+template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+__host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, GpuMat_<DstType1>& dst1, GpuMat_<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    const int rows = getRows(src1);
+    const int cols = getCols(src1);
+
+    CV_Assert( getRows(src2) == rows && getCols(src2) == cols );
+    CV_Assert( getRows(mask) == rows && getCols(mask) == cols );
+
+    dst1.create(rows, cols);
+    dst2.create(rows, cols);
+
+    grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst1), shrinkPtr(dst2), op1, op2, shrinkPtr(mask), rows, cols, StreamAccessor::getStream(stream));
+}
+
 template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType, class BinOp, class MaskPtr>
 __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType>& dst, const BinOp& op, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
@@ -134,6 +150,22 @@ __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, con
     grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst), op, shrinkPtr(mask), rows, cols, StreamAccessor::getStream(stream));
 }
 
+template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2, class MaskPtr>
+__host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType1>& dst1, const GlobPtrSz<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    const int rows = getRows(src1);
+    const int cols = getCols(src1);
+
+    CV_Assert( getRows(dst1) == rows && getCols(dst1) == cols );
+    CV_Assert( getRows(dst2) == rows && getCols(dst2) == cols );
+    CV_Assert( getRows(src2) == rows && getCols(src2) == cols );
+    CV_Assert( getRows(mask) == rows && getCols(mask) == cols );
+
+    grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst1), shrinkPtr(dst2), op1, op2, shrinkPtr(mask), rows, cols, StreamAccessor::getStream(stream));
+}
+
+
 template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType, class BinOp>
 __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, GpuMat_<DstType>& dst, const BinOp& op, Stream& stream = Stream::Null())
 {
@@ -147,6 +179,21 @@ __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, Gpu
     grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst), op, WithOutMask(), rows, cols, StreamAccessor::getStream(stream));
 }
 
+template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2>
+__host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, GpuMat_<DstType1>& dst1, GpuMat_<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, Stream& stream = Stream::Null())
+{
+    const int rows = getRows(src1);
+    const int cols = getCols(src1);
+
+    CV_Assert( getRows(src2) == rows && getCols(src2) == cols );
+
+    dst1.create(rows, cols);
+    dst2.create(rows, cols);
+
+    grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst1), shrinkPtr(dst2), op1, op2, WithOutMask(), rows, cols, StreamAccessor::getStream(stream));
+}
+
 template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType, class BinOp>
 __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType>& dst, const BinOp& op, Stream& stream = Stream::Null())
 {
@@ -159,6 +206,20 @@ __host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, con
     grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst), op, WithOutMask(), rows, cols, StreamAccessor::getStream(stream));
 }
 
+template <class Policy, class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class BinOp1, class BinOp2>
+__host__ void gridTransformBinary_(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType1>& dst1, const GlobPtrSz<DstType2>& dst2,
+                                   const BinOp1& op1, const BinOp2& op2, Stream& stream = Stream::Null())
+{
+    const int rows = getRows(src1);
+    const int cols = getCols(src1);
+
+    CV_Assert( getRows(dst1) == rows && getCols(dst1) == cols );
+    CV_Assert( getRows(dst2) == rows && getCols(dst2) == cols );
+    CV_Assert( getRows(src2) == rows && getCols(src2) == cols );
+
+    grid_transform_detail::transform_binary<Policy>(shrinkPtr(src1), shrinkPtr(src2), shrinkPtr(dst1), shrinkPtr(dst2), op1, op2, WithOutMask(), rows, cols, StreamAccessor::getStream(stream));
+}
+
 template <class Policy, class SrcPtr, typename D0, typename D1, class OpTuple, class MaskPtr>
 __host__ void gridTransformTuple_(const SrcPtr& src, const tuple< GpuMat_<D0>&, GpuMat_<D1>& >& dst, const OpTuple& op, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
@@ -449,24 +510,54 @@ __host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, GpuM
     gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst, op, mask, stream);
 }
 
+template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class Op1, class Op2, class MaskPtr>
+__host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, GpuMat_<DstType1>& dst1, GpuMat_<DstType2>& dst2,
+                                  const Op1& op1, const Op2& op2, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst1, dst2, op1, op2, mask, stream);
+}
+
 template <class SrcPtr1, class SrcPtr2, typename DstType, class Op, class MaskPtr>
 __host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType>& dst, const Op& op, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
     gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst, op, mask, stream);
 }
 
+template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class Op1, class Op2, class MaskPtr>
+__host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType2>& dst1, const GlobPtrSz<DstType2>& dst2,
+                                  const Op1& op1, const Op2& op2, const MaskPtr& mask, Stream& stream = Stream::Null())
+{
+    gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst1, dst2, op1, op2, mask, stream);
+}
+
 template <class SrcPtr1, class SrcPtr2, typename DstType, class Op>
 __host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, GpuMat_<DstType>& dst, const Op& op, Stream& stream = Stream::Null())
 {
     gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst, op, stream);
 }
 
+template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class Op1, class Op2>
+__host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2,
+                                  GpuMat_<DstType1>& dst1, GpuMat_<DstType2>& dst2,
+                                  const Op1& op1, const Op2& op2, Stream& stream = Stream::Null())
+{
+    gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst1, dst2, op1, op2, stream);
+}
+
 template <class SrcPtr1, class SrcPtr2, typename DstType, class Op>
 __host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2, const GlobPtrSz<DstType>& dst, const Op& op, Stream& stream = Stream::Null())
 {
     gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst, op, stream);
 }
 
+template <class SrcPtr1, class SrcPtr2, typename DstType1, typename DstType2, class Op1, class Op2>
+__host__ void gridTransformBinary(const SrcPtr1& src1, const SrcPtr2& src2,
+                                  const GlobPtrSz<DstType1>& dst1, const GlobPtrSz<DstType2>& dst2,
+                                  const Op1& op1, const Op2& op2, Stream& stream = Stream::Null())
+{
+    gridTransformBinary_<DefaultTransformPolicy>(src1, src2, dst1, dst2, op1, op2, stream);
+}
+
 template <class SrcPtr, typename D0, typename D1, class OpTuple, class MaskPtr>
 __host__ void gridTransformTuple(const SrcPtr& src, const tuple< GpuMat_<D0>&, GpuMat_<D1>& >& dst, const OpTuple& op, const MaskPtr& mask, Stream& stream = Stream::Null())
 {
diff --git a/modules/cudev/include/opencv2/cudev/ptr2d/glob.hpp b/modules/cudev/include/opencv2/cudev/ptr2d/glob.hpp
index 2024a7e01a2..98c115fa1bf 100644
--- a/modules/cudev/include/opencv2/cudev/ptr2d/glob.hpp
+++ b/modules/cudev/include/opencv2/cudev/ptr2d/glob.hpp
@@ -118,6 +118,18 @@ __host__ GlobPtrSz<T> globPtr(const GpuMat& mat)
     return p;
 }
 
+template <typename T>
+__host__ GlobPtrSz<T> globPtr(const GpuMat_<T>& mat)
+{
+    GlobPtrSz<T> p;
+    p.data = (T*) mat.data;
+    p.step = mat.step;
+    p.rows = mat.rows;
+    p.cols = mat.cols;
+    return p;
+}
+
+
 template <typename T> struct PtrTraits< GlobPtrSz<T> > : PtrTraitsBase<GlobPtrSz<T>, GlobPtr<T> >
 {
 };
diff --git a/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp b/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
index e68f4cf61f5..f5a3f8c85d4 100644
--- a/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
+++ b/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
@@ -49,6 +49,7 @@
 #include "../common.hpp"
 #include "../util/tuple.hpp"
 #include "traits.hpp"
+#include <cuda/std/tuple>
 
 namespace cv { namespace cudev {
 
@@ -175,4 +176,25 @@ template <class PtrTuple> struct PtrTraits< ZipPtrSz<PtrTuple> > : PtrTraitsBase
 
 }}
 
+_LIBCUDACXX_BEGIN_NAMESPACE_STD
+
+template< class... Types >
+struct tuple_size< cv::cudev::ZipPtr<tuple<Types...> > >
+: tuple_size<tuple<Types...> > { };
+
+template< class... Types >
+struct tuple_size< cv::cudev::ZipPtrSz<tuple<Types...> > >
+: tuple_size<tuple<Types...> > { };
+
+
+template<size_t N, class... Types >
+struct tuple_element<N, cv::cudev::ZipPtr<tuple<Types...> > >
+: tuple_element<N, tuple<Types...> > { };
+
+template<size_t N, class... Types >
+struct tuple_element<N, cv::cudev::ZipPtrSz<tuple<Types...> > >
+: tuple_element<N, tuple<Types...> > { };
+
+_LIBCUDACXX_END_NAMESPACE_STD
+
 #endif
diff --git a/modules/cudev/test/test_split_merge.cu b/modules/cudev/test/test_split_merge.cu
index b25c8b96d6f..598b6b80ac2 100644
--- a/modules/cudev/test/test_split_merge.cu
+++ b/modules/cudev/test/test_split_merge.cu
@@ -70,7 +70,8 @@ public:
         GpuMat_<T> d_src2(src2);
 
         GpuMat_<typename MakeVec<T, 2>::type> dst;
-        gridMerge(zipPtr(d_src1, d_src2), dst);
+        std::array<GlobPtrSz<T>, 2 > d_src = {globPtr(d_src1), globPtr(d_src2)};
+        gridMerge(d_src, dst);
 
         Mat dst_gold;
         Mat srcs[] = {src1, src2};
@@ -93,8 +94,10 @@ public:
         GpuMat_<T> d_src2(src2);
         GpuMat_<T> d_src3(src3);
 
+        std::array<GlobPtrSz<T>, 3 > d_src = {globPtr(d_src1), globPtr(d_src2), globPtr(d_src3)};
+
         GpuMat_<typename MakeVec<T, 3>::type> dst;
-        gridMerge(zipPtr(d_src1, d_src2, d_src3), dst);
+        gridMerge(d_src, dst);
 
         Mat dst_gold;
         Mat srcs[] = {src1, src2, src3};

From b236c71c2f8d983403c35a0cea8bec0432a4b0fe Mon Sep 17 00:00:00 2001
From: Peter Rekdal Khan-Sunde <peter.sunde@gmail.com>
Date: Thu, 6 Jun 2024 13:26:15 +0200
Subject: [PATCH 12/12] Merge pull request #3751 from peters:4.x

Add conditional include for <cuda/std/tuple> to support CUDA 12.4+ #3751

Fixes https://github.com/opencv/opencv_contrib/issues/3752

Adds a preprocessor check to conditionally include <cuda/std/tuple>  only for CUDA versions 12.4 and above. This ensures backward compatibility  with older legacy CUDA versions.

Related to pull request #3744

### Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

- [x] I agree to contribute to the project under Apache 2 License.
- [x] To the best of my knowledge, the proposed patch is not based on a code under GPL or another license that is incompatible with OpenCV
- [x] The PR is proposed to the proper branch
- [ ] There is a reference to the original bug report and related work
- [ ] There is accuracy test, performance test and test data in opencv_extra repository, if applicable
      Patch to opencv_extra has the same branch name.
- [ ] The feature is well documented and sample code can be built with the project CMake
---
 modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp b/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
index f5a3f8c85d4..f60ab5d0eb7 100644
--- a/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
+++ b/modules/cudev/include/opencv2/cudev/ptr2d/zip.hpp
@@ -49,7 +49,9 @@
 #include "../common.hpp"
 #include "../util/tuple.hpp"
 #include "traits.hpp"
+#if defined(__CUDACC_VER_MAJOR__) && (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 4))
 #include <cuda/std/tuple>
+#endif
 
 namespace cv { namespace cudev {
 
@@ -176,6 +178,7 @@ template <class PtrTuple> struct PtrTraits< ZipPtrSz<PtrTuple> > : PtrTraitsBase
 
 }}
 
+#if defined(__CUDACC_VER_MAJOR__) && (__CUDACC_VER_MAJOR__ > 12 || (__CUDACC_VER_MAJOR__ == 12 && __CUDACC_VER_MINOR__ >= 4))
 _LIBCUDACXX_BEGIN_NAMESPACE_STD
 
 template< class... Types >
@@ -198,3 +201,4 @@ struct tuple_element<N, cv::cudev::ZipPtrSz<tuple<Types...> > >
 _LIBCUDACXX_END_NAMESPACE_STD
 
 #endif
+#endif