Deep Learning for Face Recognition under Complex Illumination Conditions Based on Log-Gabor and LBP

Complex illumination condition is one of the most critical challenging problems for practical face recognition. In this paper, we propose a novel method based on deep learning to solve the adverse impact imposed by illumination variation in the face recognition process. Firstly, illumination preprocessing is applied to improve the adverse effects of intense illumination changes on face images. Secondly, the Log-Gabor filter is used to obtain the Log-Gabor feature images of different scales and directions, then, LBP (Local Binary Pattern) features of images sub block is extracted. Lastly, texture feature histograms are formed and input into the deep belief network (DBN) visual layer, and then face classification and recognition are completed through deep learning in DBN. Experimental results show that superior performance can be obtained in the developed approach by comparisons with some state-of-the-arts.