Verifying bank checks using deep learning and image processing

Our innovative approach completely revamps the process of verifying bank checks by leveraging deep learning and image processing. To make the cheque truncation mechanism far more efficient, we reduce the amount of human interaction. Our approach efficiently retrieves important details from the check booklet, including the bank branch code, check number, precise amount, account number, and unique signature patterns. We utilized the IDRBT cheque dataset and convolutional neural networks based on deep learning for our study. This enhanced assessment of handwritten components on bank cheques achieved 99.14 percent accuracy in identifying handwritten numeric characters. We achieved an impressive 97.7 percent accuracy while utilizing MATLAB's integrated optical character recognition (OCR) technique to interpret machine-printed text. Utilizing Scale Invariant Feature Transform (SIFT) for feature extraction and Support Vector Machine (SVM) for classification resulted in a notable 98.10 percent accuracy in signature verification.