Fusion of fingerprint presentation attacks detection and matching: a real approach from the LivDet perspective

#biometrics #recognition #fingerprint #PAD #spoofing #security

The liveness detection ability is explicitly required for current personal verification systems in many security applications. As a matter of fact, the design of any biometric verification system cannot ignore vulnerability to presentation attacks, which must be addressed by effective countermeasures from the beginning of the design process. However, despite significant improvements, especially with deep learning, current research rarely quantifies the impact of fingerprint liveness detection once integrated into full verification pipelines, mainly because practical tools to model the interaction between liveness detection and matching (both with non-zero error rates) are missing. To solve this gap, we propose the first performance simulator for sequential integration of a Presentation Attack Detector (PAD) and a matcher, by modelling probabilistic relationships between their Receiver Operating Characteristic (ROC) curves. We validate the simulator through simulations using PAD ROC results from LivDet 2017–2019 and representative matchers (NIST Bozorth3 and VeriFinger 12.0). We then use the simulator to identify PAD operating points that strengthen protection against presentation attacks without degrading the baseline verification performance.