Online Signature Verification Using Segment-to-Segment Matching
Jian Zhang, Sei‐ichiro Kamata · 2008
A new online handwritten signature verification algorithm is proposed. The proposed method divide the time functions into several segments which describing regional properties. And then a exact segment-to-segment mapping is found by a modified DTW method. Thus the proposed method utilized the dynamic features of signatures, which ensured the high security of identification systems. Furthermore, the proposed algorithm need not know any information about the characteristics of the forgeries, the decision boundary is constructed wholly according to the the obtained segment-to-segment mapping information in enrollment procedure. The proposed algorithm is tested on two separate signature verification tasks using two different signature databases, and the experimental results show affectivity of our algorithm. It also shows competitive performance of the proposed algorithm in comparison with other approaches.