Iranian Banknote Detection System

Overview

This computer vision project implements a robust image recognition system for detecting and matching Iranian banknotes using the powerful ORB (Oriented FAST and Rotated BRIEF) feature detection algorithm.

🚀 Key Features

• High Accuracy: Achieved 98.86% detection rate on test dataset
• Blazing Fast Processing: Processes 88 images in just 5 seconds (20 seconds with visualization)
• Multi-Threaded Processing: Utilizes concurrent processing for efficient image analysis
• Flexible Feature Matching: Uses ORB algorithm for reliable feature extraction
• Adaptive Image Preprocessing: Handles variations in image quality and lighting
• Generalized Design: Can be adapted for detecting various types of images, not just banknotes

📊 Performance Metrics

• Total Test Images: 88
• Correctly Detected: 87
• Accuracy: 98.86%

Why One Image Wasn't Detected?

The undetected image (500_20.jpg) is a great example of the system's realistic limitations:

This image was challenging due to:

• blur
• Poor image quality
• camera shake
• Low contrast

🛠 Installation

git clone https://github.com/yourusername/banknote-detector.git
cd banknote-detector

# Recommended: Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows, use `venv\Scripts\activate`

# Install dependencies
pip install -r requirements.txt

How It Works

1. Reference Image Loading: Loads reference banknote images from a training dataset
2. Feature Extraction: Uses ORB to detect and compute features
3. Image Preprocessing:
   • Resizes images
   • Applies Gaussian blur
   • Enhances contrast with CLAHE
4. Feature Matching:
   • Uses brute-force matcher
   • Applies ratio test for robust matching
5. Homography Estimation: Determines best image match using RANSAC

Usage

from banknote_detector import BanknoteDetector

# Initialize detector with reference images
detector = BanknoteDetector('./train')

# Detect banknote in a test image
result_image, match_index, filename = detector.detect_banknote(test_image)

Detailed Algorithm

Feature Detection

• Algorithm: ORB (Oriented FAST and Rotated BRIEF)
• Key Parameters:
  • 1000 features
  • 12 scale levels
  • Harris corner score
  • Fast threshold: 20

Matching Strategy

• Brute-force matching
• Ratio test (0.75) for robust feature correspondence
• RANSAC for outlier rejection

🌟 Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

Distributed under the MIT License. See LICENSE for more information.

Acknowledgments

• OpenCV Community
• Banknote dataset contributors
• Python Concurrent Futures library

Contact

Soroush Ebrahimi - [[email protected]]

Project Link: https://github.com/Soroushcro/banknote-detector