Abstract: Aiming at the degradation on contrast and color of single image under haze condition, the paper proposes a fast algorithm of haze removal. It analyzes the characteristics of the local mean and standard deviation on the image with fog. It obtains the difference between the local mean and standard deviation to estimate the atmospheric light intensity. It repaires the haze image combining with atmospheric scattering model. The experimental results show that the algorithm can remove the fog from the image effectively with faster processing speed and is easy to real-time applications.

Key words: haze removal, local mean, standard deviation, atmospheric scattering light, depth of scene

摘要： 针对雾天图像对比度低和颜色退化严重现象，提出一种单幅图像快速去雾算法。对雾天图像局部区域均值和标准差的特点进行分析，根据图像局部均值和标准差的差值得到关于大气散射光的估计，结合大气散射模型对雾天图像进行修复。实验结果表明，算法能够有效地去除图像中的雾气，且处理速度较快，便于实时应用。

关键词: 去雾, 局部均值, 标准差, 大气散射光, 景深