Abstract: The traditional image dehazing algorithm based on dark channel prior has some shortages such as the white border in the scene, and color distortion appeared in the bright areas of sky and cloud. In order to solve these problems, this paper proposes a fast algorithm based on color space. First, finding the minimum value in the neighbourhood is replaced by computing the mean value in solving dark channel prior. Then according to the relationship between pixel color depth in RGB space and real scene depth information, the dark colors priori is adjusted by using the weighted difference between maximum and minimum values. The threshold value is used to distinguish whether it is a bright area, and then underestimated transmittance is modified. The intensity component and saturation value are recovered in HIS color space. Finally, the histogram stretching is employed to enhance the intensity component. This method does not need image post repairing and can adaptively change the dark channel prior values. The experimental results demonstrates the effectiveness and feasibility of the proposed algorithm.