An improved dark channel dehazing algorithm is proposed to maintain the contrast constraints of the bright and dark areas of an image. According to this algorithm, the original image is first divided into light and dark two areas and the corresponding contrast ratio is calculated. Then, the dark channel dehazing algorithm based on median filtering is used to process the dark area of the image. Finally, the double histogram equalization algorithm with accurate brightness control is used to enhance the brighter area of the image with the constraint that the regional contrast constant is maximized. The results show that compared with that processed by the correlation dehazing algorithm, the final image processed by the proposed algorithm can be significantly improved in terms of information entropy, average gradient and standard deviation of brightness. The proposed algorithm can further highlight the details of the image covered by a hazy environment.