基于亮暗通道相结合的自适应图像去雾算法  

Adaptive Single Image Haze Removal UsingIntegrated Dark and Bright Channel Prior

作　　者：蒯峰阳 张丹 KUAI Feng-yang;ZHANG Dan(College of Information Science and technology,Nanjing Forestry University,Nanjing,Jiangsu 210000,China)

机构地区：[1]南京林业大学信息科学技术学院,江苏南京210000

出　　处：《计算技术与自动化》2021年第2期118-124,共7页Computing Technology and Automation

基　　金：国家自然科学基金资助项目(31170668)。

摘　　要：暗通道先验(DCP)近几年已被证实是一种合适的除雾模型,然而其过程将引起图像的Halo效应和颜色失真。基于此,提出了结合亮通道原理和天空区域分割的新算法。使用亮通道和暗通道的结合来精准估计大气光值和透射率,天空区域自适应分割解决恢复无雾图像时天空区域的色彩失真问题。将从主观及客观两方面将本文去雾算法与现有算法进行对比,结果表明,本算法能够有效消除Halo效应,获得高对比度、高色彩饱和度以及丰富细节信息的去雾结果,同时也提高了图像去雾效率。The dark channel prior(DCP)has been recently demonstrated as an adequate haze removal model.Nevertheless,its procedure will causethe Halo effect and color distortion of the image.Based on this,a new algorithm combining the bright channel principle and the sky region segmentation is proposed。It can accurately estimate the atmospheric light value and transmittance by using the combination of the light channel and the dark channel.Sky region segmentation solves the problem of color distortion in the sky region when recovering fog-free images.This article compares the defogging algorithm of this article with existing algorithms from both the subjective and objective aspects.The results show that our algorithm can effectively eliminate the Halo effect,obtain defogging results with high contrast,high color saturation and rich detailed information,improve image defogging efficiency.

关 键 词：图像去雾 亮通道 暗通道 天空分割 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]