Abstract: In view of the low accuracy of visibility estimation caused by the fixed value of haze coefficient in traditional visibility estimation algorithm. A visibility estimation method based on local contrast to optimize haze parameters. Firstly, the input image is pre-processed to solve the problem of uneven illumination. Then, the dark channel colour and atmospheric light of the image are obtained by the prior theory of dark primary colour, and the pre-processed image is segmented; the local contrast in each area block was calculated, and the haze parameters were obtained through the local contrast, and then the transmittance values corresponding to each pixel of the whole image were obtained; finally, the atmospheric extinction coefficient and visibility value can be obtained by calculating. Experimental results show that the method can achieve high precision detection within 100～500 meters, and the relative error is 15.21%. Compared with other methods, the method has higher accuracy and faster running speed.