Detection of 3D Mask in 2D Face Recognition System Using DWT and LBP

Biometrics systems have replaced traditional ways of authenticating systems. The concept of face recognition based on the fact that every person has different biological attributes. Yet, these systems have a possibility to suffer from the various types of spoofing attacks. Spoofing is the process of fooling the system by impersonating a legitimate user to gain illegal access. Face verification systems are the easiest ones to suffer from spoof attack because facial pictures are easy to access. Spoofing attack can be classified into two types: 2D and 3D. 2D attacks faked by using the photos or video of the legitimate user and 3D attacks faked by using 3D masks. Here, We have introduced a new approach to detect presence of 3D mask based face anti-spoofing using frequency and texture based feature descriptors. The proposed approach extracts Local Binary Pattern based texture features from discrete wavelet transformed images. We have evaluated the proposed approach on the 3D Mask Attack dataset that is available publicly. Our approach is performed better than state of art.