Masked Face Detection based on Micro-Texture and Frequency Analysis

Abstract

Recognition with biometric system is a vulnerable approach in security area. But face recognition approach is popular tone of them to spoofing attacks which can be done by falsifying data using craft faces and thereby gaining illegitimate access. An easy way to spoof face recognition system is to use past taken photographs instead of live person. Thus, Liveness detection is needed to design a secure system against like such illegal activities. Inspired from the facts that the images taken from the 2-D pictures and real faces have differences in characteristics like size, shape and detailedness, face prints having printing quality defects that can detected easily via using micro-texture analysis. We are proposing a novel approach analyzing texture and frequency analysis by using Local Binary Pattern (LBP) and frequency descriptor respectively. This provides a unique feature space for coupling spoofing detection and face recognition. Experiments which we were done on publicly available database produced fabulous result and we can clearly illustrate live faces and 2-D photographs.