Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Commit 8a6ea82

Browse files
cudawarpedcudawarped
committed
Replace all instances of texture references/objects with texture objects using the existing updated cv::cudev::Texture class.
Fixes bugs in cv::cuda::demosaicing, cv::cuda::resize and cv::cuda::HoughSegmentDetector.
1 parent 0792588 commit 8a6ea82

File tree

33 files changed

+1135
-2329
lines changed

33 files changed

+1135
-2329
lines changed

‎modules/cudaarithm/src/cuda/lut.cu‎

Lines changed: 13 additions & 71 deletions
Original file line numberDiff line numberDiff line change
@@ -53,15 +53,14 @@
5353
#include "opencv2/cudaarithm.hpp"
5454
#include "opencv2/cudev.hpp"
5555
#include "opencv2/core/private.cuda.hpp"
56+
#include <opencv2/cudev/ptr2d/texture.hpp>
5657

5758
using namespace cv;
5859
using namespace cv::cuda;
5960
using namespace cv::cudev;
6061

6162
namespace cv { namespace cuda {
6263

63-
texture<uchar, cudaTextureType1D, cudaReadModeElementType> texLutTable;
64-
6564
LookUpTableImpl::LookUpTableImpl(InputArray _lut)
6665
{
6766
if (_lut.kind() == _InputArray::CUDA_GPU_MAT)
@@ -73,83 +72,28 @@ namespace cv { namespace cuda {
7372
Mat h_lut = _lut.getMat();
7473
d_lut.upload(Mat(1, 256, h_lut.type(), h_lut.data));
7574
}
76-
7775
CV_Assert( d_lut.depth() == CV_8U );
7876
CV_Assert( d_lut.rows == 1 && d_lut.cols == 256 );
79-
80-
cc30 = deviceSupports(FEATURE_SET_COMPUTE_30);
81-
82-
if (cc30)
83-
{
84-
// Use the texture object
85-
cudaResourceDesc texRes;
86-
std::memset(&texRes, 0, sizeof(texRes));
87-
texRes.resType = cudaResourceTypeLinear;
88-
texRes.res.linear.devPtr = d_lut.data;
89-
texRes.res.linear.desc = cudaCreateChannelDesc<uchar>();
90-
texRes.res.linear.sizeInBytes = 256 * d_lut.channels() * sizeof(uchar);
91-
92-
cudaTextureDesc texDescr;
93-
std::memset(&texDescr, 0, sizeof(texDescr));
94-
95-
CV_CUDEV_SAFE_CALL( cudaCreateTextureObject(&texLutTableObj, &texRes, &texDescr, 0) );
96-
}
97-
else
98-
{
99-
// Use the texture reference
100-
cudaChannelFormatDesc desc = cudaCreateChannelDesc<uchar>();
101-
CV_CUDEV_SAFE_CALL( cudaBindTexture(0, &texLutTable, d_lut.data, &desc) );
102-
}
103-
}
104-
105-
LookUpTableImpl::~LookUpTableImpl()
106-
{
107-
if (cc30)
108-
{
109-
// Use the texture object
110-
cudaDestroyTextureObject(texLutTableObj);
111-
}
112-
else
113-
{
114-
// Use the texture reference
115-
cudaUnbindTexture(texLutTable);
116-
}
77+
szInBytes = 256 * d_lut.channels() * sizeof(uchar);
11778
}
11879

11980
struct LutTablePtrC1
12081
{
12182
typedef uchar value_type;
12283
typedef uchar index_type;
123-
124-
cudaTextureObject_t texLutTableObj;
125-
126-
__device__ __forceinline__ uchar operator ()(uchar, uchar x) const
127-
{
128-
#if CV_CUDEV_ARCH < 300
129-
// Use the texture reference
130-
return tex1Dfetch(texLutTable, x);
131-
#else
132-
// Use the texture object
133-
return tex1Dfetch<uchar>(texLutTableObj, x);
134-
#endif
84+
cv::cudev::TexturePtr<uchar> tex;
85+
__device__ __forceinline__ uchar operator ()(uchar, uchar x) const {
86+
return tex(x);
13587
}
13688
};
89+
13790
struct LutTablePtrC3
13891
{
13992
typedef uchar3 value_type;
14093
typedef uchar3 index_type;
141-
142-
cudaTextureObject_t texLutTableObj;
143-
144-
__device__ __forceinline__ uchar3 operator ()(const uchar3&, const uchar3& x) const
145-
{
146-
#if CV_CUDEV_ARCH < 300
147-
// Use the texture reference
148-
return make_uchar3(tex1Dfetch(texLutTable, x.x * 3), tex1Dfetch(texLutTable, x.y * 3 + 1), tex1Dfetch(texLutTable, x.z * 3 + 2));
149-
#else
150-
// Use the texture object
151-
return make_uchar3(tex1Dfetch<uchar>(texLutTableObj, x.x * 3), tex1Dfetch<uchar>(texLutTableObj, x.y * 3 + 1), tex1Dfetch<uchar>(texLutTableObj, x.z * 3 + 2));
152-
#endif
94+
cv::cudev::TexturePtr<uchar> tex;
95+
__device__ __forceinline__ uchar3 operator ()(const uchar3&, const uchar3& x) const {
96+
return make_uchar3(tex(x.x * 3), tex(x.y * 3 + 1), tex(x.z * 3 + 2));
15397
}
15498
};
15599

@@ -169,20 +113,18 @@ namespace cv { namespace cuda {
169113
{
170114
GpuMat_<uchar> src1(src.reshape(1));
171115
GpuMat_<uchar> dst1(dst.reshape(1));
172-
116+
cv::cudev::Texture<uchar> tex(szInBytes, reinterpret_cast<uchar*>(d_lut.data));
173117
LutTablePtrC1 tbl;
174-
tbl.texLutTableObj = texLutTableObj;
175-
118+
tbl.tex = TexturePtr<uchar>(tex);
176119
dst1.assign(lut_(src1, tbl), stream);
177120
}
178121
else if (lut_cn == 3)
179122
{
180123
GpuMat_<uchar3>& src3 = (GpuMat_<uchar3>&) src;
181124
GpuMat_<uchar3>& dst3 = (GpuMat_<uchar3>&) dst;
182-
125+
cv::cudev::Texture<uchar> tex(szInBytes, reinterpret_cast<uchar*>(d_lut.data));
183126
LutTablePtrC3 tbl;
184-
tbl.texLutTableObj = texLutTableObj;
185-
127+
tbl.tex = TexturePtr<uchar>(tex);
186128
dst3.assign(lut_(src3, tbl), stream);
187129
}
188130

‎modules/cudaarithm/src/lut.hpp‎

Lines changed: 1 addition & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -15,14 +15,10 @@ class LookUpTableImpl : public LookUpTable
1515
{
1616
public:
1717
LookUpTableImpl(InputArray lut);
18-
~LookUpTableImpl();
19-
2018
void transform(InputArray src, OutputArray dst, Stream& stream = Stream::Null()) CV_OVERRIDE;
21-
2219
private:
2320
GpuMat d_lut;
24-
cudaTextureObject_t texLutTableObj;
25-
bool cc30;
21+
size_t szInBytes = 0;
2622
};
2723

2824
} }

0 commit comments

Comments
 (0)