A review spoof face recognition using LBP descriptor

Abstract

Passwords are normally used for authentication in systems. They have several drawbacks like passwords can be guessed easily, they can be copied. Since biometric authentication is excelling in every field whether it be banking sector, corporate sector, etc., they are considered quite secure and mostly preferred for authentication. But every system has some flaws; therefore biometric authentication can be attacked so as to obtain any confidential information. One of them is face authentication system. Face is a unique characteristic that can be used to authenticate a person. Face authentication systems can be easily spoofed by using Replay and Printed paper attacks. Spoofing means real person’s identity is copied and used for harming any type of data. In this review paper, mainly LBP (Local Binary Pattern) descriptor is used, which is considered especially for texture analysis. LBP descriptor divides the captured face into blocks and calculates histogram for each block. Thus each block histograms are concatenated and finally are combined together. The formed histogram of whole face is compared with other face histograms and the similarity between the faces is found out. Spoof faces will not have similar histograms like the real face. And this helps in detecting Spoof face. Different spoof face detection methods are discussed in this review paper. Detection of spoof face is done by considering Moiré patterns, image distortion analysis algorithm. This review paper aims at securing confidential information by providing face unlock mechanism wherein spoof faces are to be detected.