一种基于空域占比自适应暗通道去雾算法

An Adaptive Dark Channel Dehazing Algorithm Based on Airspace Proportion

针对暗通道去雾算法存在的天空区域颜色失真的问题,提出了一种基于空域占比自适应暗通道去雾算法。该算法首先引入阈值迭代法,对有雾图像天空区域进行自适应分割;接着计算有雾图像空域占比,当占比大于设定阈值0.05时,对天空区域前1%个最高亮度像素求均值作为大气光值,否则加权平均求取全局大气光值;对于非天空区域依然按照暗通道先验计算透射率,而在天空区域引入修正因子,自适应修正天空区域透射率,同时采用引导滤波对其进行细化;最后通过大气散射模型恢复无雾图像。实验结果表明,所提算法改善了天空区域颜色失真现象,有效去除了图像雾气,恢复了图像的细节信息,且时间复杂度有所降低,具有一定优势。

Aiming at the problem of color distortion in the sky region existing in the dark channel dehazing algorithm,an adaptive dark channel dehazing algorithm based on the proportion of airspace has been proposed.Firstly,the threshold iteration method is introduced into the algorithm to adaptively segment the sky area of the foggy image;then the proportion of the airspace of the foggy image is calculated.When the proportion is larger than the set threshold of 0.05,the average value of the top 1%of the highest brightness pixels in the sky area is regarded as the atmospheric light value,otherwise the weighted average is used to obtain the global atmospheric light value;for the non-sky area,the transmittance is still calculated according to the dark channel prior,and the correction factor is introduced into the sky area,and adaptively correct the sky region transmittance while refining it with guided filtering;finally,the haze-free image is recovered by an atmospheric scattering model.The experimental results show that the proposed algorithm improves the color distortion in the sky area,which effectively removes the fog of the image and restores the detailed information of the image,and the time complexity is reduced,which has certain advantages.