Texture and quality analysis for face spoofing detection

Due to the increasing importance of security and access restrictions, the urge for implementing face recognition systems is increasing at an unprecedented rate. However, the vulnerability towards spoofing attacks limits the further progress of face recognition systems. Generally, spoofing occurs when a person attempts to gain illegal access by utilizing the biometric traits that counterfeit the registered user. Despite existence of many spoofing detection techniques in the literature, deriving a significant solution to eradicate the face spoofing attacks is quite challenging. This paper has proposed a novel method that examines the face spoofing detection system robustness with the joint exploitation of entropy-based texture and quality features. The effectiveness of this approach is evaluated by using a publicly available dataset called "replay attack database". Experimental results show the effectiveness of the proposed method in identifying various face spoofing attacks with an accuracy of 98.2% compared to existing techniques.