Dynamic Signature Verification Based on DCT of Local Features

A handwritten especially signature is the meanaccepted method to declare someone's identity, although one’ssignature may change overtime and it’s not nearly as unique ordifficult to imitate, such as fingerprints. In signatureverification field, much attention has been paid on featuresbecause verification system able to overcome its problems suchas forgeries, insensitive to intra-personal variability andsensitive to inter-personal variability. In this paper, we presenta simple and efficient approach to on-line signature verificationbased on discrete cosine transform applied to 19 time signals asposition, pressure and angles of pen. Experiments are carriedout on two benchmark databases, SVC2004 and SUSIG. Asimple method based on mean and variance of classificationerror is used to search for the best performing signals. Theproposed system is tested with different classifier for skilledforgery and the equal error rates were 5.07%, 4.33% and1.67% for SVC2004 Task1&2, Task2 and SUSIG databases,respectively.