Improved Wavelet-Based Online Signature Verification Scheme Considering Pen Scenario Information

Although signature verification is safer biotypes in terms of criminal, it has more risk on business safety. In this paper we address important challenge of on-line signature verification, which is the high dimensionality of the signature features dataset. This issue can make the verification procedure computationally costly. In this regard, we take advantage of wavelet transform along with a feature selection scheme to introduce a new set of features. The experimental results on SVC 2004 database suggest that using this set of features, we can verify the signature with 3.5% EER while having significantly lower dimension data set in comparison with the state-of-the-art techniques.