为了通过单幅真实拍摄的图像生成一系列具有雾天效果的虚拟场景,提出了一种完全基于图像的室外自然场景的雾天模拟算法.该算法首先对图像内容进行语义分割,将不同的场景内容标签为天空、地面和立于地面上的物体等;其次针对不同类型的场景分别进行深度信息的解析;最后利用大气散射模型进行雾天的模拟与仿真,从而得到室外自然场景的雾天虚拟图像.理论分析和实验结果表明,该算法的效率仅与图像的分辨率有关,且生成的雾天虚拟场景具有视觉真实感.
In this paper The algorithm takes th which are labeled as th for different model. The only relates a new algorithm is proposed to simulate ree steps. Firstly e sky, the ground regions. Finally, it theory and experime synthes nt show the foggy scene based on a real image. , it segments the input image into several semanti and the targets on the ground. Secondly, it estima izes virtual fogs in the scene using an atmospheric that the computational complexity of the proposed c regions, tes depths scattering algorithm to the resolution of the image, and the synthetic foggy scenes are visually realistic.