Dynamic Signature Verification Using Selected Regions

Abstract

AbstractIdentity verification takes into account biometric attributes. Behavioral ones are particularly important. One of them is a dynamic signature. An analysis of such type of a signature uses signals describing the signing process. In this paper, we consider the velocity signal and propose a new method for dividing the dynamic signature into groups. In this case, a group is a subset of consecutive discretization points corresponding to similar velocity values. We have also assumed here that the signature fragments characterized by the highest pen velocity are the most characteristic of each user, therefore we reject partitions related to medium and low velocity values. As a result, we individually create a unique set of partitions of different sizes for each user. We do not use skilled forgeries, which is an additional advantage of our approach. The proposed method has been tested using the BioSecure dynamic signature database. The obtained results have confirmed the effectiveness of the proposed approach.KeywordsBehavioral biometricsIdentity verificationDynamic signatureSignature partitioningSignature groups