In this project, we implement panorama stitching from scratch including corner detection, Adaptive Non-Maximal Suppression (ANMS), feature descriptor, feature matching, and RANSAC homography.
You can find the images from here.
To stitch images, you can run the following command.
python3 ./Wrapper.py --BasePath {Path to your dataset}
You can use --Dataset
to set the training set or testing set, and also --Set
to select which subset of images will be stitched.
python3 ./Wrapper.py --BasePath {Path to your dataset} --Dataset Train --Set Set1
We have shown some result in following images.
Feature Descriptors |
Feature Matching |
Feature Matching (reject ouliers by RANSAC) |
Panorama Stitching |