Offline handwritten Signature Verification using low level stroke features

Abstract

Signatures are most widely used biometric identity for verification of a person or an individual. Signature is legally accepted as a mark of identification and authorization in almost all commercial, social, jurisdictional documents since a long time. Signature verification is a process of automatic recognition of human handwritten signature and differentiating between original and forge signature. In this research, we have used low level stroke feature, which were originally proposed for recognition of printed Gujarati text, for offline handwritten signature verification. Experiment was performed on the ICDAR 2009 Signature Verification Competition dataset which contains both genuine and forge signature. Recognition is performed using Support Vector Machine (SVM) classifier with 3-fold cross validation. Equal Error Rate (EER) achieved is 15.59, which is comparable with the ICDAR 2009 Signature Verification Competition Result.