摘要： 针对雾天户外视频图像的退化现象，提出一种基于大气模型的完全自适应视频图像复原方法。该方法根据单帧图像的灰度分布特性，求出天空区域灰度的最佳近似正态分布，再由这个近似正态分布得到天空区域的灰度均值，同时根据灰度直方图分割出各个景物的深度区域并求出归一化辐射率的值，利用同深度区域内像素点的对比度实现退化图像的复原。实验表明，该方法对雾天退化图像的清晰化效果较好。

关键词: 雾天, 大气模型, 正态分布, 归一化辐射率, 深度区域

Abstract: According to degraded phenomena of outdoor video images taken in foggy weather, this paper proposes an entirely self-adapting method of video image restoration on the basis of atmospheric model. Under the gray distributing characteristic of an image, the optimal normal distribution of the gray for sky region can be gained. The gray mean of the sky distribution can be obtained by the optimal normal distribution. At the same time, the iso-depth region can also be segmented according to the gray histogram and the value of normalized radiance can be solved. It uses contrast of pixels in iso-depth region to realize restoration of degraded images. Experiments show the method can effectively improve degraded images in foggy weather.

Key words: foggy weather, atmospheric model, normal distribution, normalized radiance, depth region

中图分类号: 

• TP391.41