📷 Image Stitching Project

This is the final project for CSE 5524: Computer Vision for Human-Computer Interaction.

Description

This project implements a panoramic image stitching pipeline that combines multiple overlapping images into a seamless composite.

The core steps include:

• Feature detection and matching using SIFT (Scale-Invariant Feature Transform)

• Robust homography estimation using RANSAC (Random Sample Consensus)

• Image warping and multi-band blending for smooth transitions

Demo

1. Input images (3-view capture):

   

2. Stitched image output (before cropping):

   

3. Final cropped result:

   

Installation

1. Clone the repository to your local machine and navigate to the project directory.

   git clone https://github.com/JasonSu14/Image_Stitching.git
   cd Image_Stitching

2. Install the required dependencies.

   pip install -r requirements.txt

Usage

1. Add images you wish to stitch to the assets/inputs/ directory.

2. Modify the file paths in the notebook (image_stitching.ipynb) as needed:

   # Example
   img1_path = "assets/inputs/room_left.png"
   img2_path = "assets/inputs/room_center.png"
   img3_path = "assets/inputs/room_right.png"

3. Run all notebook cells. Final outputs will be saved to assets/outputs/.

Contributors

• Jason Su [[email protected]]
• Joe Quinn [[email protected]]