Vision Algorithms

Feature Matching and Image Stitching

Report: Link

Requirements

pip3 install opencv-python
git clone https://github.com/aditya-jha13/vision-tasks

opencv-python is installed for implementing various vision algorithms required for the task.

Installation

git clone https://github.com/aditya-jha13/vision-tasks

Usage

python3 main1.py

main1.py uses SIFT Feature Detector and BFMatcher Feature Matcher

python3 main2.py

main2.py uses ORB Feature Detector and BFMatcher Feature Matcher

Demo

Input Images

Template Image

Transformed Image

Detected Keypoints

Template Image

Transformed Image

Stitched Image

About

SIFT detectors or ORB detectors are used to detect Keypoints and their descriptions, furthermore Descriptor Matches are found using Brute Force Matcher. Then .pt attribute (Coordinates of Keypoints) are extracted from each Keypoints and a list of individual coordinates is created from matched descriptors for both Images. Finally Homography Matrix is found using those matched points and warp perspective is applied on the second Image to combine it with the First Image.

Refrences

• Keypoints: Link
• Homography: Link
• cv::warpPerspective: Link