Face liveness detection by exploring multiple scenic clues

摘要: Liveness detection is an indispensable guarantee for reliable face recognition, which has recently received enormous attention. In this paper we propose three scenic clues, are non-rigid motion, face-background consistency and imaging banding effect, to conduct accurate efficient liveness detection. Non-rigid motion clue indicates the facial motions that a genuine can exhibit such as blinking, low rank matrix decomposition based image alignment approach designed extract motion. Face-background believes of background high fake photos while faces, serve explored by GMM method. Image effect reflects quality defects introduced in reproduction, be detected wavelet decomposition. By fusing these thoroughly explore sufficient clues The proposed method achieves 100% accuracy on Idiap print-attack database best performance self-collected anti-spoofing database.