A Survey of Signature Recognition Systems: Comparative Analysis of Methods and Techniques

Abstract: This literature survey explores advancements in machine learning methodologies, specifically focusing on Artificial Neural Networks (ANN), Back Propagation Neural Networks (BPNN), and Hidden Markov Models (HMM), and their application in offline signature recognition. Highlighting key techniques, the survey reviews the use of Histogram of Oriented Gradients (HOG) and Fuzzy Min-Max Classification (FMMC), which achieve a 96% recognition rate through a diverse signature database. Additionally, it examines the Efficient Fuzzy Kohonen Clustering Network (EFKCN) algorithm, demonstrating improved accuracy in signature pattern recognition up to 70%. Emphasizing preprocessing stages, feature extraction, and robust classification frameworks, the study offers a comparative analysis of these methodologies, elucidating their theoretical foundations, practical implementations, and performance metrics.

Keywords: Artificial Neural Networks (ANN), Back Propagation Neural Networks (BPNN), and Hidden Markov Models (HMM), Histogram of Oriented Gradients (HOG) and Fuzzy Min-Max Classification (FMMC), Efficient Fuzzy Kohonen Clustering Network (EFKCN) algorithm