Facial Spoofing Detection Using Temporal Texture Co-occurrence

Biometric person recognition systems based on facial images are increasingly used in a wide range of applications. However, the potential for face spoofing attacks remains a significant challenge to the security of such systems and finding better means of detecting such presentation attacks has become a necessity. In this paper, we propose a new spoofing detection method, which is based on temporal changes in texture information. A novel temporal texture descriptor is proposed to characterise the pattern of change in a short video sequence named Temporal Co-occurrence Adjacent Local Binary Pattern (TCoALBP). Experimental results using the CASIA-FA, Replay Attack and MSU-MFSD datasets; the proposed method shows the effectiveness of the proposed technique on these challenging datasets.