Dong Ren, Junchao Chen, Chong Zhang, Zhongtu Liu, Xiaobo Liu, and Huan Zhou
Face recognition, illumination normalization, illumination qualityindex, adaptive
Illumination variation is one of the most important challenges in face recognition, because changes to illumination conditions have a significant effect on performance. However, most illumination preprocessing methods apply the same level of processing to all facial images, without considering their unique illumination conditions. Therefore, the performances of existing preprocessing methods are limited when dealing with varying illumination conditions. In this paper, we propose an adaptive illumination preprocessing method for face recognition, which adaptively preprocesses each face image according to its illumination condition. The proposed method first uses the illumination quality index (IQI) to describe the illumination. Then, we create an adaptive parameter adjustment model for the illumination preprocessing method. Finally, the proposed model adjusts the parameters of the illumination preprocessing method based on the IQI of the facial image, and enhances the preprocessing effect. Our extensive simulation results show that the proposed method can effectively improve the performance of face recognition under varying illumination conditions, when compared with existing methods.
Important Links:
Go Back
