A Unified Model for Fingerprint Authentication and Presentation Attack Detection

  • 2021-04-07 16:57:38
  • Additya Popli, Saraansh Tandon, Joshua J. Engelsma, Naoyuki Onoe, Atsushi Okubo, Anoop Namboodiri
  • 0

Abstract

Typical fingerprint recognition systems are comprised of a spoof detectionmodule and a subsequent recognition module, running one after the other. Inthis paper, we reformulate the workings of a typical fingerprint recognitionsystem. In particular, we posit that both spoof detection and fingerprintrecognition are correlated tasks. Therefore, rather than performing the twotasks separately, we propose a joint model for spoof detection and matching tosimultaneously perform both tasks without compromising the accuracy of eithertask. We demonstrate the capability of our joint model to obtain anauthentication accuracy (1:1 matching) of TAR = 100% @ FAR = 0.1% on the FVC2006 DB2A dataset while achieving a spoof detection ACE of 1.44% on the LiveDet2015 dataset, both maintaining the performance of stand-alone methods. Inpractice, this reduces the time and memory requirements of the fingerprintrecognition system by 50% and 40%, respectively; a significant advantage forrecognition systems running on resource-constrained devices and communicationchannels.

 

Quick Read (beta)

loading the full paper ...