Low illumination image Retinex enhancement algorithm based on guided filtering

It has been shown by Retinex theory that the object´s color in the image is determined by its reflectance. Under this frame work, several Retinex-based algorithms have been developed to uncover the real appearance of objects by estimating their reflectance in image. In this paper we propose an algorithm of optimizing the illumination based on guided image filter theory, where bilateral filter is utilized to estimate the illumination image. The proposed algorithm removes the halo phenomenon precisely and strengthens the partial contrast gradient effectively. Examples are given to demonstrate the effectiveness of the proposed method.