Dynamic signature verification based on DCT of local features

Abstract

A handwritten especially signature is the mean accepted method to declare someone's identity, although one's signature may change overtime and it's not nearly as unique or difficult to imitate, such as fingerprints. In signature verification field, much attention has been paid on features because verification system able to overcome its problems such as forgeries, insensitive to intra-personal variability and sensitive to inter-personal variability. In this paper, we present a simple and efficient approach to on-line signature verification based on discrete cosine transform applied to 19 time signals as position, pressure and angles of pen. Experiments are carried out on two benchmark databases, SVC2004 and SUSIG. A simple method based on mean and variance of classification error is used to search for the best performing signals. The proposed system is tested with different classifier for skilled forgery and the equal error rates were 5.07%, 4.33% and 1.67% for SVC2004 Task1&2, Task2 and SUSIG databases, respectively.