Image_stitching

A Python tool for stitching two images together using SIFT features and homography transformation.

Features

• Automatic detection of stitching direction
• Support for both horizontal and vertical stitching
• Multi-threaded feature detection
• Smooth blending at image boundaries

Implementation Details

1. Feature Detection and Matching

• Uses SIFT (Scale-Invariant Feature Transform) algorithm for feature detection
• Implements parallel processing for feature detection using ThreadPoolExecutor
• Applies ratio test for matching features between images
• Visualizes matching points in 'matching.jpg'

2. Image Registration

• Computes homography matrix using RANSAC algorithm
• Requires minimum 10 matching points for reliable transformation
• Transforms the second image to align with the first image's perspective

3. Direction Detection

• Automatically determines optimal stitching direction based on image aspect ratios
• Supports manual override for horizontal or vertical stitching
• Calculates appropriate output dimensions based on stitching direction

4. Blending

• Creates smooth transitions between images using gradient masks
• Implements weighted blending at the seam
• Automatically crops the final panorama to remove empty regions

Usage

python Image_Stitching.py '/path/to/image1.jpg' '/path/to/image2.jpg' [horizontal|vertical|auto]

Parameters

• image1.jpg: First input image
• image2.jpg: Second input image
• direction: Optional stitching direction (default: auto)
  • auto: Automatically determine direction
  • horizontal: Force horizontal stitching
  • vertical: Force vertical stitching

Output

• matching.jpg: Visualization of matched features between images
• panorama.jpg: Final stitched panorama

Limitations

• Requires overlapping regions between images
• Best results with images taken from similar viewpoints
• Performance depends on the quality of feature matches