This project provides a complete image stitching pipeline using custom-built modules for Harris corner detection, adaptive non-maximal suppression (ANMS), feature descriptors, feature matching, RANSAC, and homography transformation. The goal is to align and blend two input images based on their common features.
- Harris Corner Detection: Detects key points in the images.
- Adaptive Non-Maximal Suppression (ANMS): Refines key points to enhance the quality of the features.
- Feature Descriptors and Matching: Extracts and matches features between images.
- RANSAC: Robustly estimates the transformation matrix by fitting the model to the data.
- Homography Transformation: Aligns images based on the estimated transformation matrix.
This project utilizes several custom modules:
harris: Implements Harris corner detection.homography: Contains functions for computing homography matrices.warp: Provides image warping functionality based on homography.anms: Implements adaptive non-maximal suppression to refine key points.feature: Includes functions for finding and describing image features.ransac: Implements the RANSAC algorithm for robust model fitting.utils: Provides utility functions for plotting and saving images.
-
Clone the repository:
git clone git@github.com:z-emily/imagepano.git cd imagepano -
Install the required dependencies:
pip install -r requirements.txt
-
Run the script:
python main.py path/to/first_image.jpg path/to/second_image.jpg
Replace path/to/first_image.jpg and path/to/second_image.jpg with the actual file paths of your images.
- Results:
The processed images will be saved in the results/ directory. The results include:
harris1.jpg: Harris corners on the first image.harris2.jpg: Harris corners on the second image.anms1.jpg: ANMS points on the first image.anms2.jpg: ANMS points on the second image.matched1.jpg: Matched features on the first image.matched2.jpg: Matched features on the second image.canvas.jpg: The final panoramic image.