Offline Signature Recognition and Verification using PCA and Neural Network Approach

A Signature of a person is a unique biometric trait which can be used to validate a human identity. As signatures continue to play a vital role in legal, commercial and financial transactions, strictly secured authentication becomes more and more critical. Offline signatures are treated as the most natural means of authenticating a person’s identity. A signature by an authorized person is treated as the “seal of approval� and remains the most ideal means of authentication. This paper presents a novel approach for offline signature recognition and verification. Offline signature recognition is implemented using Principal component analysis (PCA) technique while for signature verification efficient back propagation artificial neural network is designed and trained using features extracted from Grey Level Difference Method (GLDM) technique. The system was tested against 100 test signature samples, which comprise genuine and forged signatures of ten individuals giving an average accuracy of 94%.