Off-Line Signature Verification Using Hausdorff Distance

摘要

This work proposes a new methodology for signature verification, it tries to classify, after a preliminary training, signatures between genuine and forgery. The methodology consists of finding the best compared signature disposition and then use Hausdorff distance as a similarity measure. After this, the distance calculated is used to distinguish the signatures being compared. Experiments have been done with 400 genuine signatures and 200 forgeries from 20 volunteers. The obtained results were better to signatures forgeries that the signer do not have access to the genuine signature. As an innovative approach the obtained results were very satisfactory.