Presentation Attack Detection for Face in Mobile Phones

Face is the most accessible biometric modality which can be used for identity verification in mobile phone applications, and it is vulnerable to many different presentation attacks, such as using a printed face/digital screen face to access the mobile phone. Presentation attack detection is a very critical step before feeding the face image to face recognition systems. In this chapter, we introduce a novel two-stream CNN-based approach for the presentation attack detection, by extracting the patch-based features and holistic depth maps from the face images. We also introduce a two-stream CNN v2 with model optimization, compression and a strategy of continuous updating. The CNN v2 shows great performances of both generalization and efficiency. Extensive experiments are conducted on the challenging databases (CASIA-FASD, MSU-USSA, replay attack, OULU-NPU, and SiW), with comparison to the state of the art.