利用偏振滤波的自动图像去雾

针对雾天退化图像提出一种自适应图像复原方法。 该方法基于定义的偏振图像暗通道, 自动提取图像中的天空区域, 由此获得大气光的强度和偏振度； 采用偏振滤波提取大气光强信息, 并基于最小归一化互信息原则对估计的大气光偏振度进行优化； 根据大气光强的变化规律, 对大气光强的分布进行修复； 将大气光强作为加性噪声予以扣除, 并补偿因大气衰减带来的影响, 最终复原得到场景的辐射强度信息。 实验结果表明, 该方法能够有效地改善雾天下图像的退化现象, 提高了图像的清晰度。

Automatic image dehaze using polarization filtering

To overcome the degraded images taken in hazy weather, an adaptive image restoration method was proposed. Firstly, by introducing dark channel for polarization images, sky regions are automatically segmented from the image, so the intensity and the degree of polarization of airlight can be acquired. Then, atmospheric light intensity information is extracted by polarization filtering, and used to optimize the degree of polarization of airlight by adopting the criteria of minimum normalized mutual information. After that, the distribution of atmospheric light intensity is repaired according to its change law. Finally, by removal of atmospheric light intensity, and compensation for attenuated effect of airlight, the scene radiation intensity information is recovered. Experimental results have shown that the proposed algorithm can alleviate the degradation of the image efficiently, and enhance the definition of the image.