Face Spoofing Detection using Multiscale Local Binary Pattern Approach

摘要

Face spoofing is one of the problem faced in face authentication system. Nowadays devices contain patterns and passwords for logging into the systems but they are prone to some disadvantages. There are other biometric traits like finger, iris etc. But extra devices are used for such detail capturing of the biometric trait. Face authentication does not require any extra hardware for authenticating a person. Spoof attacks include Replay attack and Printed paper attack, where Printed paper attack involves presenting printed photo of the authenticated user in front of the camera. The motive of this paper is to detect such spoofing attacks on the face authentication systems used in desktop. Nanjing University of Aeronautics and Astronautics (NUAA) photograph imposter database consisting of 15 samples of Printed photo attacks are used for further testing of the proposed system. Currently MLBP and SIFT histogram plotting of the captured face and spoof is obtained, this histogram will be considered for classifying the face as spoof or not.