结合双通道与光源定位的夜间图像去雾  

Night Image Dehazing Combining Dual-Channel and Light Source Positioning

作　　者：陈奕涵 刘奇[2] 何柯辰 陈曦 CHEN Yi-han;LIU Qi;HE Ke-chen;CHEN Xi(College of Electrical Engineering,Sichuan University,Chengdu Sichuan 610065,China;College of Biomedical Engineering,Sichuan University,Chengdu Sichuan 610065,China)

机构地区：[1]四川大学电气工程学院,四川成都610065 [2]四川大学生物医学工程学院,四川成都610065

出　　处：《计算机仿真》2023年第12期220-225,共6页Computer Simulation

摘　　要：针对夜间雾图光照不均、色偏严重等问题,提出了一种基于亮暗双通道与光源定位的夜间图像去雾方法。结合夜间雾天成像模型,首先对图像进行超像素分割,根据评分阈值区分远近光源区域,对远光源区域采用双通道先验原理自适应确定大气光值,再对景深一致的超像素块进行最小滤波操作求取透射率,并通过引导滤波进行透射率的优化,最后针对光源变化引起的色偏问题,采用基于标准差加权的灰度世界算法进行颜色校正。与经典去雾方法相比,上述方法在对比度、平均梯度、信息熵等指标上有较大提升,验证了上述去雾方法的有效性。Aiming at the problems of uneven illumination and serious color cast in nighttime fog images,this pa⁃per proposes a nighttime image dehazing method based on bright and dark dual channels and light source positioning.Combined with the nighttime foggy imaging model,the image was first segmented by superpixels,and the near and far light source areas were distinguished according to the score threshold.For the far light source area,the two-chan⁃nel prior principle was used to adaptively determine the atmospheric light value.The minimum filtering operation was performed on the superpixel blocks with the same depth of field to obtain the transmittance,and the transmittance was optimized by guided filtering.Finally,aiming at the problem of color shift caused by the change of light source,the gray world algorithm based on standard deviation weighting was used for color correction.Compared with the classic dehazing methods,the method in this paper has a great improvement in contrast,average gradient and information entropy,which verifies the effectiveness of the dehazing method in this paper.

关 键 词：夜间图像去雾 亮暗通道先验 光源定位 超像素分割 颜色校正 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]