🌄 Panorama Stitching

An advanced panorama stitching pipeline built with Python and OpenCV. It stitches multiple overlapping images into a seamless wide-angle panorama using cylindrical projection, feature matching, homography estimation, seam blending, and automatic horizon correction.

✨ Features

• Cylindrical projection to reduce perspective distortion
• SIFT feature detection and descriptor extraction
• Feature matching using BFMatcher
• RANSAC-based homography estimation with affine fallback
• Global exposure compensation for brightness balancing
• Seam finding for smooth transitions
• Multi-band blending using Laplacian pyramids
• Automatic horizon/tilt correction
• Debug output images for feature match visualization

📁 Project Structure

Panaroma-stitching/ │ ├── main.py # Entry point and CLI handling ├── stitcher.py # Main panorama stitching pipeline ├── utils.py # Helper functions and image utilities ├── projection.py # Cylindrical projection logic ├── features.py # Feature detection and matching ├── homography.py # Homography estimation ├── warper.py # Image warping and canvas generation ├── exposure.py # Exposure compensation ├── seam_blend.py # Seam finding and blending └── README.md

⚙️ Installation

Clone the repository and install dependencies:

git clone https://github.com/jasman5/Panaroma-stitching.git cd Panaroma-stitching

pip install opencv-python numpy scipy

🚀 Usage

Basic usage:

python main.py --input img1.jpg img2.jpg img3.jpg --output panorama.jpg

Available Arguments

Argument	Required	Description
--input	✅	Input images in left-to-right order
--output	✅	Output panorama image path
--focal	❌	Focal length for cylindrical projection
--no-cylinder	❌	Disable cylindrical projection

📌 Examples

Basic Stitching

python main.py --input left.jpg center.jpg right.jpg --output pano.jpg

Using Custom Focal Length

python main.py --input img1.jpg img2.jpg --output pano.jpg --focal 800

Disable Cylindrical Projection

python main.py --input img1.jpg img2.jpg --output pano.jpg --no-cylinder

🔍 Stitching Pipeline

Input Images ↓ Cylindrical Projection ↓ Feature Detection (SIFT) ↓ Feature Matching ↓ Homography Estimation (RANSAC) ↓ Warping onto Common Canvas ↓ Exposure Compensation ↓ Seam Finding & Blending ↓ Auto Crop & Horizon Correction ↓ Final Panorama

📷 Tips for Better Results

• Capture images left to right
• Keep 30–50% overlap between images
• Lock camera exposure and focus
• Avoid moving objects during capture
• Rotate around a fixed point for minimal distortion

🛠️ Tech Stack

• Python 3.x
• OpenCV (cv2)
• NumPy
• SciPy Jasman Kaur — GitHub Profile