AutoPano - Panorama Stitching Project

This project demonstrates a panorama stitching pipeline, combining multiple images into a seamless panoramic image.

Pipeline Overview

1. SIFT Feature Detection

The Scale-Invariant Feature Transform (SIFT) is used to detect and describe local features in the images. SIFT is robust to changes in scale, rotation, and illumination, making it ideal for identifying key points across overlapping images.

2. Feature Matching with RANSAC

The detected SIFT features are matched between images to find corresponding points. RANSAC (Random Sample Consensus) is applied to remove outliers and ensure that only the best matches are used for further processing. This step ensures the alignment of images is accurate and robust.

3. Warping Images

Using the matched feature points, images are geometrically transformed and aligned to create a continuous view. This involves computing homographies that warp each image so that overlapping areas are aligned correctly.

4. Blending with Distance Transformation

The aligned images are blended using distance transformation techniques to ensure smooth transitions and minimize visible seams. This method helps in achieving a coherent panoramic image with minimal artifacts.

Results

• 5 images

• 3 images

• 8 images