Offline Signature Verification Using Local Interest Points and Descriptors

Abstract

In this article, a new approach to offline signature verification, based on a general-purpose wide baseline matching methodology, is proposed. Instead of detecting and matching geometric, signature-dependent features, as it is usually done, in the proposed approach local interest points are detected in the signature images, then local descriptors are computed in the neighborhood of these points, and afterwards these descriptors are compared using local and global matching procedures. The final verification is carried out using a Bayes classifier. It is important to remark that the local interest points do not correspond to any signaturedependent fiducial point, but to local maxima in a scale-space representation of the signature images. The proposed system is validated using the GPDS signature database, where it achieves a FRR of 16.4% and a FAR of 14.2%.