musaibahmedrazzaqui / cheque-it

Cheque It Fraud Detction System

This was made in part of a Research Collaboartion of Polaris Technologies with our Technical Advisors.

This project implements a signature forgery detection and cheque verification system using a combination of image processing and deep learning techniques. The project's goal is to provide a reliable solution for detecting forged signatures and verifying bank cheques.

Technical Summary

Data Collection

• Signatures: Genuine and forged signatures data collected from Kaggle and CEDAR datasets.
• Cheques: High-resolution bank cheque images collected from Kaggle.

Data Preparation

• Signatures: Data organized into real and forged folders. Image processing includes rotations, rescaling, and horizontal flips using Keras Image Generator.
• Cheques: Preprocessing involves BGR to grayscale conversion, thresholding, edge detection, dilation, erosion, and noise reduction using OpenCV.

Model Training

• Signatures: Convolutional Neural Networks (CNNs) used with Keras and TensorFlow. The model includes 6 layers with convolution, batch normalization, max pooling, dropout, and fully connected layers.

Image Segmentation

• Signatures: Images segmented to extract relevant parts (amount and signature) using OpenCV functions.

Testing and Evaluation

• Model accuracy validated using a holdout set. Data augmentation and additional layers employed to improve accuracy.
• Cheque verification includes user-submitted image and account number, with the system detecting forgery or errors.

Conclusion

• Successful implementation demonstrated for signature forgery detection and cheque verification.
• High accuracy achieved on Kaggle and CEDAR datasets.
• Proposed system contributes to Industry Innovation and Infrastructure SDG by automating document verification.
• System improves efficiency in detecting and investigating forgeries, simplifying authentication for bank staff.

Purpose

The project's purpose is to provide a reliable and practical solution for authentication and security in financial processes. By combining data collection, image processing, and deep learning, the system effectively detects signature forgery and verifies bank cheques.