I have implemented the Panoramic image stitching using invariant features from scratch. Implemented the David Lowe paper on "Image stitching using Invariant features".
NOTE: You can experiment with any images (your own choice). I have experimented with many images. You can check result below. You can find many images in "Image_Data" folder.
CREATE DATA:
- You can create multiple images like S1.jpg, S2.jpg, S3.jpg, S4.jpg and S5.jpg (shown below) .Make sure there will be some overlapping parts between two consecutive created images in a sequence. then only algorithm will find and match features and create panorama image of all images which you have provided.
- OR you can directly feed multiple images from camera in a sequence with some overlapping parts between two consecutive images.
Please install Libraries:
- Numpy
- opencv-contrib-python
TO RUN CODE:
-
Put images in a folder .
-
Enter the image name in order of left to right in way of concantenation. Like: Enter the 1 image: S1.jpeg Enter the 2 image: S2.jpeg
-
Run stitch.py code.
python stitch.py --image <image folder path> --output <output folder path>- Then, you will get your panorama image as stitch.jpeg in your image dataset folder.
- Used SIFT to detect feature and then RANSAC, compute Homography and matched points and warp prespective to get final panoramic image.
RESULTS:
Result of S1.jpeg, S2.jpeg, S3.jpeg, S4.jpeg and S5.jpeg