Design of anti-spoofing system for face recognition

Abstract

Of the many types of cyber-security measures today, biometrics stands out as one of the most intriguing, with great potential to be explored. The advancement of technology means that threats and the ways one can attack is increasing. Face spoofing detection is one way to overcome the attack of spoofing. However, the current methods of anti-spoofing via facial recognition utilises only the light (luminance) aspect of images, without paying much attention to the chrominance or colour part. This report therefore seeks to implement and prove the functionality of a new methodology- one that involves the heavy emphasis of chrominance, through colour texture analysis. This is done using the Michigan State University’s Mobile Face Spoofing Database (MSU MFSD) and MATLAB. The results of these experiments are then compared and concluded against its literature derivation. At the end, the potential for using Deep Learning Methodology is introduced.