AUTOMATICALLY STATIC SIGNATURE VERIFICATION USING ANGULAR FEATURE AND DIRECTION

Abstract

Signature Verification is a extensively and commonly used mechanism for authentication of an individual because of social and legal acceptance and massive use of the written signature. There are two categories of signature verification based on the acquisition of the viz., on-line and off-line verification system. Off-line signature verification is more challenging task of biometric than on-line signature verification because the features are extracted from the static 2D image of the signature therefore behavioral properties of the signature image is absent. In this thesis, an approach based on chain code histogram features and angular features is proposed for off-line signature verification. In the proposed approach 8-direction chain code his-togram of each grid on the contour signature image is extracted. Angular features are extracted in two phases. In the first phase geometric center skeleton signature is used to extract the features and in the second phase input signature image is divided into fixed size grids to extract angular features. The extracted features from all the signature image constitutes the knowledge base. The SVM (Support Verification Machine) is used as verification tool. SVM is trained with the randomly selected training samples. Extensive experimentation have been conducted to exhibit the performance of the proposed approach on publicly available dataset, CEDAR. A comparative study is done to justify the feasibility of the proposed approach for off-line signature verification over the existing approaches.