Supervised Neural Network for Offline Forgery Detection of Handwritten Signature
Summra Saleem, M. Ghani Usman, Aslam Muhammad, Martinez-Enriquez A.M. · 2021 18th International Conference on Electrical Engineering, Computing Science and Automatic Control (CCE) · 2021
Authorized high security systems for legal contracts is need of the hour for an automatic authentication system. This study investigates the off-line signature verification system to identify skilled forgery based on Writer-Independent system approach. This research advances our understanding of layer architecture in deep convolutional neural network (CNNs) to perform specific task using optimized learning parameters. Four layered architecture with Adam optimizer produces significant accuracy of 88.39% for self-generated off-line handwritten signatures. The proposed model efficiently classifies forged signatures from genuine ones.