Fingerprint skeleton matching based on local descriptor

Abstract

In this paper, we present a new fingerprint matching algorithm based on a local skeleton descriptor. This descriptor uses ridge count information to encode minutiae locations in a small neighborhood. Taking advantage of ridge count properties, our descriptor is robust to distortions. We developed an efficient algorithm to match our descriptor and a strategy to combine matchings of many local descriptors. Our algorithm obtains interesting results on both tenprint-to-tenprint and latent-to-tenprint matchings.