Minutiae Selection using Reference Point for Fingerprint Data Interoperability and Identification

The most crucial features of fingerprint recognition are minutiae, which are friction ridge ending and bifurcation. Given an input fingerprint, minutiae are extracted and encoded into a template for exchange between two organizations. The number of minutiae is limited for the fixed size template, especially when embedding minutiae into a smart card. In this work, we propose a minutiae selection method that can choose a required number of most reliable minutiae from all extracted minutiae of the input fingerprint. These robust minutiae are selected based on distance from a reference point and a corresponding minutiae density. We benchmark the proposed method against three previous techniques with two fingerprint matching algorithms on the NIST SD14 database. The experimental results show that the proposed method achieves most of the best performance in terms of the rank-1 identification rate for both matching algorithms.