结合暗通道与图像分层的图像去雨雾方法  

Image Rain and Fog Removal Method Combining Dark Channel and Image Layering

作　　者：张爱苓 李少荣 张瑞 

机构地区：[1]西安石油大学电子工程学院,陕西西安710065

出　　处：《工业控制计算机》2022年第12期67-69,共3页Industrial Control Computer

摘　　要：针对目前大部分算法只能解决单一天气下的图像增强问题,提出一种结合暗通道先验和引导滤波进行图像分层的图像去雨去雾算法。首先对图像进行暗通道先验,大气光值的计算选择亮通道与暗通道结合的方法,之后用引导滤波对透射率进行细化,得到经暗通道先验处理后的图像,再将图像进行引导滤波分层,对低频图像进行锐化,减少低频细节的损失,将高频图像进行多次引导滤波,获得更多高频图像中的背景分量,最后将低频图像与多次引导滤波后的无雨高频图像叠加。经实验表明,该算法相较于其他算法,既可以完成有雾图像的增强,又能实现有雨图像的复原,且在峰值信噪比和结构相似性上也均有较大提升,实现了复杂天气下的图像增强。Aiming at the fact that most of the current algorithms can only solve the problem of image enhancement in a single weather,a rain and fog removal algorithm based on image layering combined with dark channel prior and guided filtering is proposed in this paper.Firstly,the dark channel priori is performed on the image.The method of combining bright channel with dark channel is selected for the calculation of atmospheric light value.Then,the transmittance is refined by guided filtering to obtain the image processed by dark channel priori.Then the image is layered by guided filtering to sharpen the low-frequency image and reduce the loss of low-frequency details.The high-frequency image is guided and filtered for many times to obtain more background components in the high-frequency image.Then the low-frequency image is linearly superimposed with the high-frequency image after multiple guided filtering.

关 键 词：图像去雨雾 暗通道先验 引导滤波 透射率细化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]