Fingerprint generation and presentation attack detection using deep neural networks

초록

Performance evaluation of ?ngerprint recognition systems requires large-scale databases. Unfortunately, collecting ?ngerprints is expensive and time-consuming, and publishing them is restricted due to the privacy protection legislation. Hence, an algorithm which can generate huge ?ngerprint datasets would be of great help. With the popularization of ?ngerprint authentication systems, detecting fake ?ngerprints, also known as presentation attack detection, is an essential problem. Inspired by the fast development of deep learning, this paper demonstrates novel algorithms to generate arti?cial ?ngerprints and detect fake ?ngerprints using deep neural networks. The experimental results prove that the proposed system can generate ?ngerprints which have the same characteristics as real ?ngerprints. Regarding presentation attack detection, the proposed system shows an average detection error rate of 1.57% on three LivDet databases, including LivDet 2011, 2013, and 2015.