消除光晕效应和保持细节信息的图像快速去雾算法

现有的基于大气散射物理模型的图像去雾算法，在去雾过程中大都无法避免的会产生光晕效应和细节丢失。针对这一问题，提出了一种消除光晕效应和保持细节信息的图像快速去雾算法。首先运用四叉树子矩阵划分的分层遍历方法得到更精确的大气光值，再通过分析大气散耗函数，利用融合梯度信息的改进引导滤波得到精确估计的大气散耗函数，并自适应的获取最小值图像与大气光平均值的阈值，求解出透射图，最后反演复原出无雾图像，并对复原后的图像进行亮度调整。对多组有雾图像进行了实验，本算法能有效地抑制去雾过程中产生的光晕效应，较多地保留了图像的细节信息，且运行时间大约减少了一倍。融合梯度信息的改进引导滤波不但可以较好地保留透射图的细节信息，有效地消除光晕效应，而且具有较好的鲁棒性和时间复杂性。本算法适用于交通等室外场景的去雾。

Fast algorithm for image defogging by eliminating halo effect and preserving details

Objective: The existing image defogging algorithm based on atmospheric scattering model, the defogging process are inevitable will have a halo effect and details are lost. To solve this problem, this paper proposes a fast defogging algorithm for eliminating halo effect and keep the detail information of the image. Method: The algorithm first uses four tree sub space division level the search method to get more accurate atmospheric light value, then consumption function through the analysis of atmospheric dispersion, using the improved gradient information fusion filter to get the accurate estimation of atmospheric lead consumption dispersion function, and adaptive image and get the minimum average value of atmospheric optical threshold, calculated the transmission map, finally no fog inversion recovery image, and the brightness of the restored image. Result: Experiments on fog images are carried out. The algorithm can effectively suppress the halo effect in fog removal. More details are preserved and the running time is reduced by about twice. Conclusion: Retain detail information transmission improved fusion gradient information guide filter not only can better eliminate the halo effect, effective, and has better robustness and time complexity. This algorithm is suitable for transportation and other outdoor scenes to defog.