Fingerprint Liveness Detection Based on Pore Analysis

Fingerprint scanners can be spoofed by using artificial fingerprints made from Play-Doh, gelatin and silicone molds. It is, thus, necessary to offer protection for fingerprint systems against such threats. Since it is difficult to replicate pores during the fake fingerprints fabrication, in this paper, we propose a pore based liveness detection method. In our method, we firstly extract the sweat pores from a grayscale fingerprint image using the Mexh wavelet transform and adaptive Gaussian filters. Based on the obtained pore information, we compute five statistical features of the image to quantify the pore distribution. Finally, the support vector machine (SVM) technique is used for classification. Our experimental results show the proposed method can achieve an average classification error rate (ACE) of 7.11%, 11.4% on ATVS-FFp and LiveDet2011 data sets, respectively, generally outperforming conventional liveness detection schemes to be compared.