For the images captured in the bad weather like fog or haze, the atmospheric scattering effect not only seriously affects the visual appearance of the image, but also hinders the image feature extraction. Therefore, it needs de-fog technology for image enhancement and restoration, to improve the visual effects and convenience of post-processing. Because that dark colors prior algorithm is time consuming and has poor treatment effect etc., we put forward an improved algorithm of boundary constraints defogging algorithms. At the same time, we introduce information entropy and average gradient to evaluate the algorithm objectively. Comparison of experimental results shows that this method has a high computing speed, and better effects on the deal.