This project implements an image stitching pipeline in Python using OpenCV.
It merges two overlapping images into a single panoramic view by detecting, matching, and aligning keypoints.
The system compares two feature detection methods:
- Harris Corner Detector (with patch-based descriptors)
- SIFT (Scale-Invariant Feature Transform) for scale and rotation invariance
A custom RANSAC-based affine transformation estimator is used to filter outliers and achieve accurate alignment.
Finally, a distance-weighted blending method ensures smooth transitions between overlapping regions.
- Harris and SIFT keypoint detection
- Brute-force descriptor matching
- Custom RANSAC for affine transformation
- Image warping and blending
- Sensitivity analysis (thresholds & top matches)
- Python
- OpenCV
- NumPy
- Matplotlib