On-line signature verification based on dynamic Bayesian network

In this paper, we present a novel approach to automatic on-line signature verification using dynamic Bayesian network (DBN). On-line signatures can be viewed as time-series data. For modeling time-series data, it's natural to use directed graphical models, which can capture the fact that time flows forward. The model is called dynamic Bayesian network and it hasn't been used in online signature verification. Experimental evaluation on a database containing a total of 3500 signatures of 100 individuals shows promising results compared to other classical methods.