Off-line signature verification based on a new symbolic representation and dynamic time warping

Abstract

Handwritten signature verification is a classical technology. Which uses features extracted from signatures. The purpose of signature verification is to identify whether a testing signature image is the same as a template. In this study, a new off-line signature verification method based on symbolic representation and dynamic time warping is proposed. In the proposed symbolic representation, the distances between symbols are defined according to the average value of each segment rather than the distance calculated based on Gaussian distribution, which is used in traditional symbolic representations. The proposed dynamic time warping (DTW) method is used to detect the optimal path and an optimal verification result can then be obtained by comparing the fitting errors of all matching paths. To demonstrate the performance the proposed approach, the following experiment is carried out: firstly, a signature image was created by a person and this is used as the template. Secondly, a total of 100 signature images written by the same person were collected, and the false rejected rate (FRR) and false acceptance rate (FAR) are calculated using the proposed method. Thirdly, a total of 100 signature images written by 100 different people were collected, based on which the false acceptance rate is calculated. The experimental results show that the performance of the proposed method is very promising, and the FRR and FAR values for the two group tests are 3% and 2% respectively.