Date of Publication :20th November 2023

Abstract: Handwritten signature recognition, a pivotal component of biometric authentication, demands robust and efficient feature extraction techniques for optimal performance. This research presents a comparative analysis of three prominent feature extraction methods: Local Binary Patterns (LBP), Histogram of Oriented Gradients (HOG), and Scale-Invariant Feature Transform (SIFT). Using a curated dataset of 2,000 signatures, comprising both genuine instances and skilled forgeries, we evaluated each technique's efficacy in terms of accuracy, computational efficiency, and robustness. Our findings revealed that while HOG demonstrated superior accuracy, LBP excelled in computational speed, and SIFT showcased potential in handling varied capture scenarios. This study provides valuable insights for the development of advanced signature recognition systems, emphasizing the significance of tailored feature extraction for enhanced biometric authentication.

Reference :