Video-based Signature Verification By Tracking Pen Tip Using Particle Filter Combined With Template Matching

摘要

A signature verification system based on video is proposed in this paper. The signature imformation is obtained from video of camera which is different from traditional online or offline method. First, we set the camera in upper front of the white written paper, so that we can get the signatures have the most identifiability for human recognition. We locate the pen tip position using a contour-based method. And then we track the pen tip using a particle filter and template matching method. At last, we classify a unknown signaure as genuine or forged signature by considering the distance between the unknown signature and the referenced signature with a threshold. We performed an experiment on self-built database consisting of 160 signatures from 20 subjects. The experimental results show that our method can track the nonlinear-moved pen tip eaxctly, can be used for most common pens, and this system has more potential capability for signature verification.