An Improved Dominant Point Feature for Online Signature Verification

Among the biometric characteristic, signature forgery is the easiest way to do. Possibility of signature forgery similarity might be reached perfectly. This paper introduced a new technique to improve dominant point feature system based on its location for online signature verification. Dynamic Time Warping is used to match two signature features vector. The performance of system is tested by using 50 participants. Based on simulation result, system accuracy without presence of the simple and trained impostors is 99.65% with rejection error is 0% and acceptance error is 0.35%. While the current systems are faced with the simple and trained impostors, system accuracy became 91.04% with rejection error is 1.6% and an average of acceptance error is 7.36% with details as follows; acceptance error is 0.08%, acceptance error of simple impostors is 4.4%, and acceptance error of trained impostors is 17.6%. The improved feature within fusion is produce better accuracy significantly than dominant point feature. Accuracy of the improved feature within fusion is 91.04%, whereas system accuracy with just use the dominant point feature is 70.96%.