本文提出了两种基于雾气理论的视频去雾新算法,一种视雾气为覆盖在各帧图像上的一层遮罩被从原视频中减除,一种视雾气为光路传播图被分离消除.前者将Retinex算法得到的亮度图像与视频帧自身的深度关系相结合求取雾气遮罩,并将此遮罩从原视频帧中分离以去除雾气;后者将由背景图像得到的视频“通用”传播图应用于视频的所有帧以消除雾气.实验证明两种算法均能有效地提高原有雾视频各帧的对比度和清晰度.本文所提算法从雾气的角度人手,无需借助参考图像,运算代价低,与一般视频去雾算法相比,在获得较优的去雾效果的同时,具有较好的实用性和较快的处理速度.%In this paper,we proposed two new video defogging algorithms based on fog theory. One is regarding fog as the veil layer to be subtracted, and the other is taking fog as the transmission map to be separated from the original video. The former uses the luminance component image obtained by Retinex algorithm and the depth information of the original video frames to separate the veil layer. The latter applies a single transmission map obtained from the background image to a series of video frames. Experiments show that both algorithms can effectively improve the contrast and quality of the video frames. Compared with other state of the art algorithms,our algorithms restore video frames from a perspective of fog with no reference image and low computation cost.The new algorithms can remove fog effectively as well as provide a good practicability and a fast speed.
展开▼
