Novel Algorithm for the On-Line Signature Verification

On-line signature is a biometric attribute used in a identity verification process. One of the most effective methods of signature verification is the method based on partitioning of signature trajectories. In this paper a concept of new approach to identity verification based on partitioning of trajectories is presented. In this approach signature is partitioned into subspaces which are weighted by weights of importance. The weights are used in classification process. Partitions associated with high values of weight have greater importance in classification process than partitions associated with low weight values. The algorithm was tested with use of public on-line signature database SVC 2004.