A Novel Illumination Normalization Algorithm

This paper studies the illumination normalization of images which are captured under complex lighting condition and analyses the problems of halo and gradient reversal exist in traditional techniques theoretically. Based on the weighted least square filter (WLS), this paper proposes a new illumination normalization algorithm. The proposed algorithm gets the accurate estimation of the illumination of the image by using the weighted least square filter and then eliminates the effects brought by light changing, it also enhances the contrast adaptively by using the modified S function. Experiment results show that the proposed algorithm can eliminates the degradation caused by light changing drastically and improves the post processing algorithms´ performances effectively.