The application of the Bayes Ying Yang harmony based GMMs in on-line signature verification

摘要

In this contribution, a Bayes Ying Yang(BYY) harmony based approach foron-line signature verification is presented. In the proposed method, a simplebut effective Gaussian Mixture Models(GMMs) is used to represent for eachuser's signature model based on the prior information collected. Different fromthe early works, in this paper, we use the Bayes Ying Yang machine combinedwith the harmony function to achieve Automatic Model Selection(AMS) during theparameter learning for the GMMs, so that a better approximation of the usermodel is assured. Experiments on a database from the First InternationalSignature Verification Competition(SVC 2004) confirm that this combinedalgorithm yields quite satisfactory results.