Abstract:

A method is presented to create a polar coordinate system based on the mass center of whole signature and the mass center of golden section under which  the extrema can be extracted from polar angle feature as signature segmentationpoint sequence. This article proposes an improved DTW algorithm for online handwritten signature verification that is applied to the
 segments matching between the signature sample and the template automatically, obtaining the highest qualified corresponding relationship of the segments and the most closely similarity measurement result, which can degrade the FRR level efficiently. The algorithm has been tested via the signature data from the handwritten signature database JLUDHSDB2.0 and the experiment results show that the EER is 4.25 percent.

Key words: online handwritten signature verification, dynamic time warping (DTW), pattern recognition, handwritten signature database