Pseudoletter Heightened Overlapped Handwritten Cheque Segmen-tation and Forgery Detection Framework Using Sgcip-Cnn in Banking

Abstract

Segmentation of handwritten Bank Cheque Images (BCI) helps to extract important details and makes smoother transactions. However, in previous works, overlapped Handwritten Texts (HT) with Printed Texts (PT) are not concentrated, disrupting the segmentation process. Therefore, this framework used Intelligent Character Recognition (ICR) and Pseudo Letter with Height-based Segmentation (PLHS) to effectively remove the overlapped texts for accurate HT segmentation. At first, the BCI is preprocessed to obtain a denoised enhanced image. From the enhanced BCI, the overlapped HT with PTs are recognized and segmented separately using ICR and PHLS. Then, the date, signature, name, and amount from HT are separately identified using Nanonets. After that, the faded texts are redrawn using texture inpainting followed by edge detection and contour formation. Finally, the recognized texts are categorized as forged and genuine cheques using a Sigmoidal Growing Cosine Intermap Pooling-based Convolutional Neural Network (SGCIP-CNN) with a classification accuracy of 98.7899% for effective money transactions. Thus, the proposed work outperformed the existing methodologies.