Offline signature verification using geodesic derivative pattern

In this paper, Geodesic Derivative Pattern (GDP) for Off-line handwritten signature verification is presented. We combine features based on both gray level and geometric information in the decision level. The Local Derivative Pattern (LDerivP) and the geodesic distance are used as features. It should be mention that the geodesic distance has never been used in offline signature verification. The method is tested on the GPDS960GraySignature database. Just one genuine sample per person has been used to train a KNN model and the remaining samples have been used for testing. Experimental evaluation demonstrates that the Geodesic Derivative Pattern (GDP) performs much better than the LDeriveP for offline signature verification.