Towards minimizing effect of partial fingerprint images on the performance of fingerprint recognition systems

Partial fingerprint matching is still a challenging problem to be solved. Modern devices like smartphones, laptops, PDAs etc are equipped with compact silicon based fingerprint sensors which may produce partial fingerprints. In this paper, we present partial fingerprint problem as quality issue. Performance of the fingerprint recognition system heavily depends on the quality of the fingerprint image. We evaluated NIST´s NFIQ quality estimator as a performance predictor for Bozorth3 matching algorithm. We proposed new Quality estimator NFIQCore which is an extension of NFIQ with addition of core point feature. Experimental evaluation of proposed quality estimator is performed on FVC 2002 Db 1a. Replacing NFIQ with NFIQcore shows significant improvement on Equal Error Rate (EER) from 3.30 to 0.87.