Application of HMM to online signature verification based on segment differences

Abstract

In this paper, a novel application of Hidden Markov Model (HMM) to online signature verification is proposed, this application utilizes segment difference values obtained by segmentation Dynamic Time Warping (DTW) as observations of HMM. It combines the advantages of segmentation DTW which measures the features in local, and advantages of HMM which models the variability of observation sequences in global. Firstly, correspondences of the critical points in signatures are marked by segmentation DTW. Then, a variety of differences between corresponding segments are calculated by classical DTW. Finally, HMM is trained by utilizing these differences. In this paper, the practical meaning of the model states is clear and can be illustrated as degrees of similarity. Consequently, the HMM topology is set to ergodic. The validity of the proposed method was tested on the public SVC2004 signature database. © Springer International Publishing 2013.