Robust Hybrid Texture Descriptor (HTD) and a parallel score based fusion for face verification and liveness detection system

Abstract

Currently, most of the biometric recognition systems are based on face verification are susceptible to the spoof attacks. Video replays, printed photographs and 3D mask attacks provoke false acceptance lest some necessary counter-measures are employed. We focus on still face spoof attacks considered as one of the most easily generated attacks and challenging for modern face verification systems. To detect face spoofing, most of existing countermeasures focus on designing discriminative features to analyze the textural properties of facial skin. To improve the texture discriminating properties and get advantages from other texture descriptor, in this paper, a novel Hybrid Texture Descriptor (HTD) is proposed and models the joint performance of face verification and liveness detection by score fusion method, for which the multi-modeling is well recognized. Numbers of experiments are carried out on UPM face spoof and one public domain Replay attack databases as standalone countermeasure in addition to the consolidation with verification system by means of fusion of score. The potential of proposed method demonstrate the outperformance throughout the experiments.