Improved face recognition method under variable lighting conditions

Zhaosheng Liu,Bo Li,Xiuping Liu,Zhixun Su

2009-01-01

Journal of Computational Information Systems

Abstract:This paper presents a new face recognition method under variable lighting conditions. The proposed method has the ability to factorize a single face image and obtain the illumination invariant facial structure, which is then used for face recognition. Compared with the traditional illumination normalization methods, our method has better edge-preserving ability. In the face recognition step, we present a topology preserving nonlinear mapping method. The key idea is to construct a nonlinear function, which maps the original training space to a feature space. The feature space should have the similar topology structure to the original space, and should be more distinctive. Finally, the results of experiments show that our proposed method is more stable to variable illuminations. Copyright © 2009 Binary Information Press.