基于区间估计的快速单幅图像去雾方法

单幅图像去雾是一个病态反问题。为了解决这个问题，对于大气光强度和介质传输率的估计已进行了大量研究。本文从新的角度提出一种简单但有效的基于区间估计的单幅图像快速去雾方法。该方法从大气散射模型出发，首先，通过对大气光强度和介质传输率取值范围的讨论，减小区间估计的范围。其次，通过快速联合双边滤波和中值滤波对介质传输率进行边缘优化，并利用大气光强度与暗通道差值绝对值大小对介质传输率中明亮区域进行判断和修正。最后，通过色调调整对复原图像进行整体优化。实验结果表明，与几种典型的图像去雾方法比较，本文方法明显提高了算法运算速度，并获得了较好的复原效果。

Fast single image dehazing method based on interval estimation

Single image dehazing is an ill-posed problem of the image degradation model. Many researches have been done to solve the problem in estimating the global atmospheric light and the medium transmission. In this paper, a simple but effective interval estimation method is proposed to quickly remove haze from a single input image in a new perspective. Firstly, a prior discussion is used to reduce the value scope of global atmospheric light and medium transmission. Secondly, the fast joint bilateral filtering and median filtering are adopted to optimize the edge regions of medium transmission. And by the absolute value of the difference of atmospheric light and dark channel, the bright areas can be judged and amended. Finally, the tone mapping is used to refine the performance of the restored image. Experimental results demonstrate that the proposed method can clearly improve the computational speed and obtain a better restoration effect compared to some state-of-the-art methods.