Fingerprint Presentation Attack Detection Approaches in Open-Set and Closed-Set Scenario
Anshdha Gupta, Akhilesh Verma · Journal of Physics Conference Series · 2021
Abstract In the biometric system, fingerprints are widely used to recognise an individual’s identity and are vulnerable to presentation attacks such as spoofing. Spoof fingerprints are fake fingerprints created using artificial materials like play-do, silicone, etc., which fool the biometric identification system. Spoof handling is an urgent need nowadays, as these attacks’ success rate is more than 70%. Much work has been done on fingerprint spoof detection, but mostly involved closed-set problems, i.e. spoof materials used are the same for training and testing set, but they face poor generalisation problem. Hence to overcome closed-set problems, Open-set solution came into existence in which some novel spoof materials were introduced in the testing set, which was not seen during training. This paper will give a brief review of various closed-set and open-set solution techniques along with the tabular view of their Average Classification Error (ACE) scores.