Project Proposal
Title
Currency Recognition and Fake Note Detection using MATLAB
Abstract
Circulation of counterfeit currency is a growing issue that damages the economy and financial
systems. Manual detection methods are unreliable and time-consuming. This project proposes a
MATLAB-based automated system that uses image processing techniques to differentiate
between real and fake currency. The system captures a currency note image using a camera,
processes it to extract security features (such as watermark, security thread, and texture), and
classifies the note as genuine or counterfeit. The solution is cost-effective, real-time, and useful
for banks, retail shops, and ATMs.
Objectives
1. To acquire currency note images using a camera/webcam.
2. To preprocess images for enhancement and noise removal.
3. To extract key security features such as watermark, security thread, and texture.
4. To classify the note as Real or Fake using MATLAB.
5. To provide an automated and reliable solution to detect counterfeit currency.
Problem Statement
Counterfeit currency circulation is increasing rapidly and poses serious challenges for financial
transactions. Manual verification is prone to human error and cannot guarantee reliability.
Therefore, an automated detection system based on image processing is required to identify fake
notes accurately and efficiently.
Project Components
Hardware Components
High-resolution Camera/Webcam (for image capture).
Proper Lighting setup (to highlight hidden features).
Original and counterfeit currency samples (for training/testing).
Software Components
MATLAB (main development environment).
Image Acquisition Toolbox (for capturing live images).
Image Processing Toolbox (for preprocessing and feature extraction).
Computer Vision Toolbox (for feature detection and pattern matching).
Deep Learning Toolbox (optional, for CNN-based classification).
Methodology
1. Image Acquisition
o Capture image using webcam.
o Functions: webcam, snapshot.
2. Preprocessing
o Convert to grayscale (rgb2gray).
o Resize image (imresize).
o Remove noise (medfilt2).
o Adjust contrast (imadjust).
3. Feature Extraction
o Watermark detection: Intensity analysis.
o Security thread detection: Edge detection (edge).
o Texture features: GLCM (graycomatrix, graycoprops).
o Keypoint features: SURF (detectSURFFeatures, extractFeatures).
4. Classification
o Machine learning models:
SVM (fitcsvm)
KNN (fitcknn)
o Optional: Deep learning CNN (trainNetwork).
5. Decision Making
o Compare extracted features with reference dataset.
o If similarity > threshold → Note is Real.
o Otherwise → Note is Fake.
Tools and MATLAB Features
webcam, snapshot – Image acquisition
rgb2gray, imresize, medfilt2, imadjust – Preprocessing
edge, graycomatrix, graycoprops – Feature extraction
detectSURFFeatures, extractFeatures – Keypoint detection
fitcsvm, fitcknn, trainNetwork – Classification
Expected Outcomes
A reliable system to identify real vs. fake currency.
Real-time detection capability.
Accuracy improvement compared to manual verification.
Practical use in banks, ATMs, shops, and exchange centers.
Applications
Banking sector for automated verification.
ATMs for real-time note validation.
Retail and business shops.
Currency exchange and financial institutions.
Conclusion
This project applies image processing and machine learning techniques in MATLAB to detect
counterfeit currency efficiently. By analyzing watermark, texture, and security thread features,
the system will provide an accurate and real-time solution, reducing reliance on manual detection
and helping prevent counterfeit circulation.