一种基于降维滤波的快速去雾算法

A Fast Dehazing Algorithm Based on Dimension Reduction Filter

为了达到单幅图像的快速去雾,提出了一种基于降维滤波的快速去雾算法.采用边界约束来求取透射率图,然后用降维滤波对透射率图进行优化,减少了计算量;通过自适应调节大气光值和透射率下限值优化了天空区域的失真;对去雾后的图像进行自适应Zadeh-X变换改善图像的视觉效果.通过对多种带雾图像实验表明该算法可以达到高速和高质量的去雾,并且适用于偏暗、偏亮和颜色不丰富的带雾场景.

In order to remove fog for single image fleetly,a defogging method based on dimension reduction filter is proposed in this paper.It uses boundary constraint to achieve transmission map,and then optimize the transmission map with dimension reduction filter.The complex computation is reduced considerably.Distortion of the image in the sky area is compensated by adaptively adjusting the atmospheric optical value and lower limit of transmission map.Then the visual effect of optimized image is improved by applying an adaptive Zadeh-X transformation.Experimental results on a variety of haze images demonstrate real-time performance and high-quality of the proposed algorithm,and it is applicable to foggy scenarios despite illumination is too low or too high and the color information is not rich enough