基于物理模型的单幅雾天图像去雾方法

雾天图像对比度降低，为了保证图像的可见性，需要对图像进行去雾处理。根据光学原理，雾天条件下场景的能见度下降是与场景深度呈指数关系的。为了对雾天图像进行清晰化，提出一种基于大气光学物理模型的图像去雾算法。该算法首先用四叉树方法估计出雾天图像的环境光；然后估计出使输出图像对比度最大化的传输函数；最后进行雾天图像复原。试验表明，该方法是有效的和有用的。

Single Fog Image Defogged Method Based on Physics Model

As the fog image contrast reduction, we need to defog the image in order to ensure the visibility of the image. According to optics theory, the reduction of scene visibility is exponential to the scene depth. This paper proposes an image defogging algorithm based on physical model of atmospheric optics. First, the proposed algorithm estimates the ambient light in a given hazy image based on the quad-tree subdivision. Then, the proposed algorithm estimates the transmission map to maximize the contrast of the output image. Finally, image restores in the fog. The experiments demonstrate the effectiveness and practicality of the proposed algorithm.