Due to the high-bright area may lead to a bad impact on the estimation of ambient light,and the color disto rtion m ay happen in that area,which will influence the result of restoration,an improved algorithm is proposed for single image in this paper.Fir stly,the bright area and the dark area are separated into two categories based on fuzzy classification.The ambient light is estimated using da rk channel prior theory based on these bright areas to improve the accuracy.Because the bilateral filter not only has the feature of smoothing,but also has the ability to sustain the edge details o f image,the atmospheric dissipation function can be estimated through a bilatera l filter,and then we can get the atmospheric transmittance,which will be more accurate and refined c ompared with traditional methods.At last,the brightness and tone of color distortio n zone are adjusted based on the human vision to improve the quality of the whole image restored.Compared with so me classic algorithms,the time complexity and space complexity of the algorithm are decreased,while details and color fidelity are better in bright area.Experimental results show that the prop osed algorithm is effective and feasible.