On-line Chinese signature verification based on the partition algorithm

摘要

In this paper, we present a novel method used for on-line handwriting signature verification, which relies on the speed and pressure partition of the signature curves. The velocity and pressure of the reference signature is divided into two groups, such as high or low velocity partitions. These partitions of a genuine signature possess reasonable stability. We selected the most stable one of all partitions by means of statistical computation, which is used for setting the decision boundary, and obtained the signature template. According to the signature template, we used the principal component analysis (PCA) method to rotate the test signature, and align the length of the test signature relying on dynamic time warping (DTW). The experimental results show that this method is effective and achieves a false rejection rate (FRR) of 10.7%, a false acceptance rate (FAR) of 0.8% and an average error rate (AER) of 5.8%, respectively.