This repository is created to perform homography transformation between two images, including shifting, rescaling and rotation.
Can be used in many cases, while we focus more on a specific case of only cropping(shifting) from original image and then rescaling.
Required config(can be changed in config.py):
- ROOT_PATH: root path of the input and output images
- SEARCH_RADIUS: radius of searcing for best start and end point
- MIN_MATCH_PAIRS: min match pairs needed for calculation
Required: python3, cv2
Output:
query_keypoints.jpg: keypoints of query image
train_keypoints.jpg: keypoints of train image
features_matching.jpg: features matching and localization of query and train image
result.jpg: calculate query image from train image, should be the same as the old query image
How to run:
- Edit config.py for any paths/settings
- run
python main.py
