Optimized Classification Approach For Offline Signature Verification System

Now a day, security applications and person's identity verification using biometric systems is increasing rapidly for security aspects. In this field of security system, banking field also requires a proper authentication for transaction when any transaction is done through cheque. In that case, signature matching becomes a challenging task due to skilled forgery in signatures. Hence, a technique is required for offline signature verification systems. Generally, signatures may vary each time which may mislead to the verification process. To overcome this issue, we have presented an improved classification approach for offline signature verification system where simulated annealing optimization scheme is implemented for optimized trainset formulation by computing best cross-validation scores. In order to show the performance, we have considered GPDS &CEDAR database are considered where skilled forgery is present. Experimental study shows better performance when compared with other state of art techniques.