中文摘要：

目前,基于暗通道先验的去雾算法是目前最有效的单幅图像去雾方法之一.文中研究比较了各种基于暗通道先验的雾化图像去雾技术;由于采用软抠图算法的暗通道先验去雾算法时间复杂度太大,文中提出利用自适应中值滤波实现对透射率的平滑与细化,以提升去雾算法的速度;提出了一种具有鲁棒性的大气光值求取方法.实验结果表明,文中所提方法能实现雾化图像的有效去雾,能保证去雾图像的清晰度,大幅提高暗通道先验去雾算法的速度,具有一定实用价值.

英文摘要：

Currently, defogging algorithm based on dark channel prior is one of the most effective single image defogging method. This paper studies various image defogging technologies based on dark-channel-prior. The time complexity of soft matting algorithm is too large. This paper presents an adaptive median filter to achieve a smooth and refined transmittance to improve the algorithm speed and a robust method for determining the value of Light Atmosphere. Experimental results show that the proposed method can achieve effective defogging, ensure defogging image clarity, and increase speed of dark-channel-prior defogging algorithm, which has some practical value.