🧠 VisionNote

A Computer Vision System for Currency Authentication

---

💡 Overview

VisionNote is a Reliable Artificial Intelligence project designed to authenticate Indian currency notes (₹500 & ₹2000) using computer vision and image processing techniques.
The system applies structured visual analysis, feature extraction, and similarity comparison to determine whether a currency note is genuine or counterfeit, with strong emphasis on accuracy, transparency, and explainability.

It has been built completely in Python using:

• OpenCV for image processing
• Tkinter for GUI development
• SSIM and ORB algorithms for feature matching
• Jupyter Notebook as the development environment

---

🧩 Key Features

• 🔍 Detects counterfeit Indian currency of ₹500 and ₹2000 denominations
• ⚙️ Utilizes ORB (Oriented FAST and Rotated BRIEF) for key feature detection
• 📊 Measures similarity using SSIM (Structural Similarity Index) for comparison
• 🖥️ User-friendly Tkinter GUI for interaction and visual results
• 📁 Includes a custom, structured dataset of real and fake notes
• 🧠 Built with Reliable AI principles — transparent, reliable, and explainable

---

🛠️ Libraries and Tools

Library / Tool	Purpose
OpenCV	Image processing and core feature extraction
Tkinter	Graphical User Interface (GUI) for input and output
NumPy	High-performance numerical operations and array handling
Matplotlib	Visualization and plotting of data and features
Jupyter Notebook	Modular development, testing, and control flow

---

📂 Project Structure

VisionNote/
│
├── Dataset/
│ ├── Real_Notes/ # Real ₹500 and ₹2000 notes for templates
│ ├── Fake_Notes/ # Fake currency note images for testing
│ └── Features/ # Stored security feature templates
│
├── Fake Notes/ # Sample fake notes for testing
├── 500_testing.ipynb # Notebook for ₹500 detection logic
├── 2000_testing.ipynb # Notebook for ₹2000 detection logic
├── controller.ipynb # Main notebook controlling the workflow and GUI launch
├── gui_1.ipynb # GUI module for user input (Image selection, denomination)
├── gui_2.ipynb # GUI module for displaying detailed results
├── VISIONNOTE_REPORT.pdf # Complete project report
└── README.md # You are here!

---

⚙️ How to Run

Step 1: Setup and Initialization

1. Clone the repository:

   git clone [https://github.com/your-username/VisionNote.git](https://github.com/your-username/VisionNote.git)
   cd VisionNote

2. Open the project in Jupyter Notebook:

   jupyter notebook

3. Run the main notebook:
   • Open controller.ipynb
   • Click Run All cells.

---

Step 2: Use the GUI

1. A GUI window (gui_1.ipynb) will launch.
2. Click Select an Image and choose a note image (sample images are in the Dataset/ folder).
3. Select the correct denomination (₹500 or ₹2000).
4. Click Submit.

Step 3: View the Result

1. The system processes the image (~5 seconds).
2. A new GUI window (gui_2.ipynb) displays the detailed authenticity report, scores, and final label.

---

📊 Results and Analysis

Category	Notes Tested	Correctly Classified	Accuracy
Real Notes (₹500 & ₹2000)	19	15	79%
Fake Notes	12	10	83%

⏱️ Average Processing Time: ~5 seconds per note

🧾 Decision Rule: If $\geq 9$ out of 10 security features pass the SSIM/count checks, the Note is classified as Genuine.

---

🔒 Reliable AI Principles

Principle	Implementation in Project
Transparency	The system displays SSIM scores for each feature, showing the precise metrics used for the decision.
Explainability	Uses interpretable, rule-based computer vision metrics (SSIM, contour counts) instead of opaque black-box ML models.
Reliability	Built with deterministic algorithms and tested against a verified, custom dataset.
Accessibility	Designed with a user-friendly GUI for non-technical operators.
Fairness	Avoids human/data bias through consistent, objective, rule-based image analysis.

---

🖼️ Demo (Screenshots)

To give a visual overview of the user experience:

1. Image Upload and Input Window

2. Processing Screen

3. Final Results Screen

---

🧭 Future Enhancements

• 🏦 Expand Denominations: Extend support to include other notes (₹10, ₹20, ₹50, ₹100, ₹200).
• 🤖 Integrate Machine Learning: Implement deep learning (e.g., CNNs) for automatic feature localization and recognition to improve robustness.
• 📱 Deployment: Develop mobile and web-based versions for broader public and institutional use.
• ☁️ Cloud Validation: Introduce cloud-based template verification for real-time validation and scalability.

---

🏁 Conclusion

VisionNote: A Computer Vision System for Currency Authentication effectively demonstrates the practical application of Trustworthy Artificial Intelligence principles, including accuracy, interpretability, and accessibility, to address a real-world financial security challenge. By integrating structured computer vision techniques with ethical, rule-based AI methods, the system ensures transparent and explainable counterfeit detection. The project delivers a reliable, scalable, and user-centric solution that highlights how responsible AI and computer vision can be combined to build trustworthy systems for critical applications such as currency authentication.