Abstract: In the haze and fog climatic condition, the quality of the captured pictures will be reduced for the reason of atmospheric attenuation. An improved method based on median filter is proposed to recover the chroma of scene and improve the visual effect. Firstly, according to the characteristics of the one-dimension entropy and average gray value of input image, self-adaptively regulating of the global atmospheric light can achieve the optimal brightness of haze-free image. Furthermore, to reconcile the haze-free image chroma, the image fusion method is introduced into this paper, and the whole optimization of dehazing system is realized based on the fusion parameter proportion set by the value of global atmospheric light. Experimental results show that the dehazed images with the proposed method have better visual effect. The chroma, brightness and constrast have been improved.

Key words: dehazing, median filter, self-adaptive, image fusion

摘要： 在雾、霾气候下所捕获到的照片质量因大气衰减而有所降低。为有效恢复场景实际色度，提高图像的视觉效果，提出了一种基于中值滤波去雾的改进算法。首先根据输入图像的一维熵以及平均灰度的特点，自适应调节大气光成分值使得去雾后图像的亮度达到最优；为调和去雾后图像的色度，引进了图像融合的方法，并且根据大气光成分值的选取特点设置了融合比例，从而实现了去雾系统的整体最优化。实验结果表明，算法去雾后图像具有较好的视觉效果，其色度、亮度、对比度都有了一定程度的提高。

关键词: 去雾, 中值滤波, 自适应, 图像融合