Face liveness detection based on frequency and micro-texture analysis

Facial biometric system is a widely used approach in security industry. But face recognition systems are vulnerable to spoofing attacks which can be done by falsifying data using non-real faces and thereby gaining illegal access. An easy way to spoof face recognition systems is to use portrait photographs instead of the real person. Thus, Liveness detection is needed to make a system secure against such spoofing attacks. Inspired from the fact that the images taken from 2-D photographs and live faces are bound to have differences in terms of shape and detailedness, we present an approach based on frequency analysis and texture analysis by using frequency descriptor and Local Binary Pattern (LBP) respectively. Experiments which were done on publicly available database showed excellent results and can efficiently classify live faces and 2-D photographs.