Off-line signature verification using geometric features specific to Chinese handwriting

Abstract

This paper is concerned with the off-line signature verification scheme. The distinction in our work is that we have taken more consideration with the Chinese signature structure. And we present four main features for the optimization of the verification of the Chinese signatures, viz, the envelop of the signature, cross-count feature, center of gravity of sub-region and distance between vectors made of center of gravity, and area of embedded white space. Experimental results show that the combination of the four-feature based classifiers increases the verification accuracy, particularly for the Chinese signature verification.