Colorful image enhancement algorithm based on guided filter and Retinex

Retinex is a theory on the human visual perception. It explains an object's color is determined by its reflectance. The classic Retinex causes the problems of halo, color distortion and noise, etc. In order to solve these problems, In this paper, we present a algorithm that combining with Nonlinear Image Enhancement and Guided Filter to estimate the reflection components under Retinex frame work. The Nonlinear Image Enhancement Algorithm enhances the details and the Guided Filter optimizes the illumination of the input images. We demonstrate the algorithmn is both effective and efficient in better color recover and contrast, also it removes the halo and noise phenomenon precisely.