🎬 Video Stabilization & Comparison Toolkit (Python Version)

This project is a Python implementation of Lakshya-Kejriwal/Real-Time-Video-Stabilization.
The original algorithm is described in the paper: A Hybrid Filtering Approach of Digital Video Stabilization for UAV Using Kalman and Low Pass Filter.

---

🚀 Project Overview

A powerful, easy-to-use toolkit for video stabilization and multi-algorithm comparison, based on Python and OpenCV. Supports single video stabilization, batch comparison, interactive GUI comparison, and more. Perfect for de-shaking, algorithm evaluation, and academic research.

---

📁 Directory Structure

video-stable-v1/
│
├── main.py                        # Main entry, single video stabilization example
├── video_stabilizer.py            # Video stabilization algorithm (Kalman filter + affine transform)
│
├── compare/
│   ├── batch_comparison.py        # Batch video comparison tool (CLI)
│   ├── interactive_comparison.py  # Interactive comparison tool (Tkinter GUI)
│   └── video_stabilization_comparison.py # Core class for multi-video comparison
│
├── comparison_output.mp4          # Example comparison output
├── stabilized_output_kalman.avi   # Example stabilized output
├── stabilized_output.avi          # Other output example
└── ...

---

🛠️ Requirements

• Python 3.7+
• OpenCV (opencv-python)
• numpy
• tkinter (standard library, required for GUI)

Install dependencies:

pip install opencv-python numpy

---

✨ Features

• Real-time video stabilization using Kalman filter and affine transformation
• Batch video comparison for algorithm evaluation
• Interactive GUI for visual comparison and result export
• Easy integration and extensibility for research and development
• Feature point detection algorithm selectable: Supports 'gftt' (default), 'orb', 'sift', and can choose different feature point extraction methods based on requirements.

---

📦 Module Descriptions

main.py

• Single video stabilization demo
• Supports webcam or local video file input
• Outputs to stabilized_output_kalman.avi
• Feature type selectable: Set feature_type to 'gftt' (default), 'orb', or 'sift' in the code to choose the feature detection algorithm.

video_stabilizer.py

• Core stabilization algorithm (feature tracking + affine + Kalman filter)
• Automatic cropping and resolution recovery

compare/batch_comparison.py

• Command-line batch comparison tool
• Compares all videos in a folder and generates a composite video

compare/interactive_comparison.py

• Tkinter-based interactive GUI
• Drag-and-drop, order adjustment, real-time preview, export

compare/video_stabilization_comparison.py

• Core class for multi-video comparison
• Frame sync, label display, grid stitching, output saving

---

🚦 Usage

1. Single Video Stabilization

Edit main.py to set your input video path and select the feature detection algorithm:

feature_type = "gftt"  # Options: "gftt" (default), "orb", "sift"
stabilizer = VideoStabilizer(feature_type=feature_type)

Then run:

python main.py

Output: stabilized_output_kalman.avi

2. Batch Video Comparison

python compare/batch_comparison.py

Follow the prompt to enter the video folder path. A comparison video will be generated automatically.

3. Interactive Comparison (Recommended)

python compare/interactive_comparison.py

A GUI will pop up. Add multiple videos, set the output path, and click "Start Comparison".

---

❓ FAQ

• Black border issue: The algorithm does not automatically remove all black edges. Some cropping is performed, but black borders may still appear depending on the transformation.
• Resolution/frame rate mismatch: The comparison tool adapts automatically, but it is recommended that input videos have the same parameters.
• Missing dependencies: Please ensure all required libraries are installed.

---

🤝 Acknowledgements

• Original C++ Project: Lakshya-Kejriwal/Real-Time-Video-Stabilization
• Original Paper: Real Time Video Stabilization Using Kalman Filter

---

📬 Contact & Contribution

Feel free to submit Issues or Pull Requests for suggestions, bug reports, or contributions!

---

If you need a more detailed technical description or a Chinese version, please let me know!