Automatic online signature verification based on dynamic function features

Abstract

Online signature verification has been a research topic of great interest, in line with the increasingly high security requirements of a networked society. In this paper, an online signature verification system based on the nearest template matching is designed. A set of function features are employed for comparing the dissimilarity between the test signature and the template database. The difference in verification performance is investigated between using a commonly applied distance threshold for all signers and using a signer-dependent distance threshold. Two online signature databases (Chinese and Subcorpus-100 MCYT) are used for evaluation. Good verification results are achieved for both databases. The achieved equal error rate (EER) is 1.69% and 1.77% for Chinese and Subcorpus-100 MCYT, respectively. In addition, a full comparison with the most relevant state-of-the-art methods is provided.