KNN and FKNN Based Modified DTW on Dynamic Signature Verification

KNN and FKNN Based Modified DTW on Dynamic Signature Verification

Signature verification is a reliable and publicly acceptance method for authentication. The efficiency of any signature verification system depends mainly on thediscrimination power and robustness of the features use in the system. This paper evaluates 16 dynamic features viewpoint classification error and discrimination capability betweengenuine and forgery signatures. A modified distance of DTW algorithm is proposed to improve performance of verificationphase. The proposed system is evaluated on the public SVC2004 signature database. The experimental results show that first, themost discriminate and consistent features are velocity-based. Second, average EER for proposed algorithm in comparison with the general DTW algorithm show a relative decrease 50.4%.Moreover, comparative study based on KNN and FKNN classifiers with skilled forgery show that FKNN has lower errorrates.