Local Features Analysis of On-Line Signature Using Modified Distance of DTW

Abstract

Signature verification is a reliable and publicly acceptable method for authentication. Each signature is represented as a set of functional features such as coordinates of signature points, pen pressure and pen angle and therefore many features are available to the designer of signature verification system. The efficiency of any signature verification system depends mainly on the discrimination power and robustness of the features use in the system. This paper evaluates 40 dynamic features viewpoint classification error and consistency for extracting the best subset once a set of features provide maximal discrimination capability between genuine and forgery signatures. A modified distance of dynamic time warping (DTW) algorithm is proposed to improve performance of verification phase. The proposed system is evaluated on the public SVC2004 signature database. The experimental results show that first, the most discriminate and consistent features are velocity-based. Second, average equal error rate (EER) for proposed algorithm in comparison with the general DTW algorithm show a 47.5% decrease. Moreover, comparative study based on different classifier with skilled forgery show that the best result has EER of 1.73% using Parzen window classifier.