Abstract

This review examines the use of Support Vector Machine (SVM) classification for automated handwritten signature verification, a vital aspect of biometric authentication and fraud detection. Addressing challenges like intra-class variations and inter-class similarities, the system leverages MATLAB's SVM framework with image preprocessing techniques, including dimensional standardization (100x100 pixels), grayscale conversion, and vector transformation for feature extraction. Using a linear kernel, the SVM effectively constructs optimal hyperplanes to distinguish genuine signatures from forgeries with high accuracy and computational efficiency. The system excels in handling variations in signature styles and writing pressure, demonstrating robust performance through empirical testing and comparisons with recent studies. Future improvements could explore advanced feature extraction, alternative kernel functions, and deep learning integration to enhance reliability in real-world applications. This review underscores SVM's effectiveness as a reliable tool for biometric authentication.