Under the bad weather, scattering of atmospheric particles lead to the degradation of image quality. And then the later image threshold segmentation is affected. We propose a moving-window threshold segmentation algorithm based on contrast enhancement. According to the characteristics of gray levels and by way of different histogram enhancement, the image contrast can be effectively improved. Moving-window threshold segmentation can reconstruct image gray space. In accordance with the certain rules and artificial selection of a small piece of Am XAn, threshold segmentation of sub-block can be done. Then the threshold segmentation of the whole image can be obtained through progressive scan from top to bottom. Then, by combining the split result together and smoothing the image block adjacent joint, the final image segmentation is obtained. The experimental results show that image gray histogram completely enhances the haze image and efficiently restrains the noise.