A new method for signature verification based on selection of the most important partitions of the dynamic signature

Identity verification based on the dynamic signature is an important issue of biometrics. Analysis of the signature dynamics, especially selecting partitions in order to improve processing precision, is very often used to evaluate whether a given signature is genuine. In our previous works we have presented methods for selecting partitions with a specific interpretation. They correspond to low and high velocity of the pen and low and high pressure of the pen in the initial, middle and final time phases of a signing process. In this paper we propose a new partitioning method which selects a specific, i.e. characteristic set of partitions for each user individually. The purpose of this approach is: (a) to increase the precision of signature processing, (b) to individually adapt the method to the specifics of the user signatures by eliminating redundant partitions, and (c) to simplify test signatures verification. The tests of the proposed method were performed using the BioSecure signature database.