Abstract

A prevalent method for personal identification and validation, handwritten signatures find extensive use in various settings. Documents such as bank checks and legal agreements necessitate signature authentication, presenting a formidable challenge given the sheer volume. This underscores the need for a robust automated signature verification tool to combat fraud across financial transactions. Presently, manual verification relies heavily on the verifier's subjective judgment, which is prone to inefficiencies and errors. Expert evaluators struggle to discern the subtle differences in line and angle ratios between genuine and fraudulent signatures. To address these challenges, we propose an automated signature verification system leveraging advancements in image processing and machine learning. Upon capturing a customer's handwritten signature, preprocessing steps will enhance its clarity and extract relevant features. Subsequently, a verification process will compare these features against stored data in the database. Our approach employs OCR-based techniques for signature localization, combined with Connected Components and geometric feature analysis. Classification using Support Vector Machine (SVM) achieves a commendable 91% accuracy in signature authenticity verification.