Abstract: In this paper,a new algorithm is proposed to simulate the foggy scene based on a real image.The algorithm takes three steps.Firstly,it segments the input image into several semantic regions,which are labeled as the sky,the ground and the targets on the ground.Secondly,it estimates depths for different regions.Finally,it synthesizes virtual fogs in the scene using an atmospheric scattering model.The theory and experiment show that the computational complexity of the proposed algorithm only relates to the resolution of the image,and the synthetic foggy scenes are visually realistic.