舰船电子工程2024,Vol.44Issue(2) :49-54.DOI:10.3969/j.issn.1672-9730.2024.02.011

基于天空分割和暗通道先验的图像去雾算法

Image Defogging Algorithm Based on Sky Segmentation and Dark Channel Prior

甄国涌 张伟康 单彦虎 储成群
舰船电子工程2024,Vol.44Issue(2) :49-54.DOI:10.3969/j.issn.1672-9730.2024.02.011

基于天空分割和暗通道先验的图像去雾算法

Image Defogging Algorithm Based on Sky Segmentation and Dark Channel Prior

甄国涌 1张伟康 1单彦虎 1储成群1
扫码查看

作者信息

  • 1. 中北大学仪器与电子学院 太原 030051
  • 折叠

摘要

针对传统暗通道先验算法对天空区域去雾时出现的颜色失真等问题,提出一种基于天空分割和加权计算大气光值的去雾算法.首先基于有雾图像中天空区域梯度小、对比度低和亮度高的特征实现天空区域的分割.其次基于天空分割的结果加权计算大气光值,有效避免了图像中小面积高亮区域对大气光值预估的影响,防止恢复的图像偏暗.最后通过对不同区域透射率的结合,实现图像的复原.实验结果表明:该算法能够对包含天空区域的有雾图像有效复原,解决了天空区域颜色失真的问题,且去雾后的图像细节清晰、颜色自然.

Abstract

Aiming at the problems such as color distortion when the traditional dark channel a priori algorithm defogs the sky region,a defogging algorithm based on sky segmentation and weighted calculation of atmospheric light value is proposed.Firstly,the segmentation of the sky region is realized based on the features of small gradient,low contrast and high brightness of the sky re-gion in the foggy image.Secondly,the atmospheric light value is weighted based on the result of sky segmentation,which effectively avoids the influence of small bright areas on the atmospheric light value prediction and prevents the recovered image from being dark.Finally the image recovery is realized by combining the transmittance of different regions.The experimental results show that the algorithm can effectively recover the foggy image containing the sky region,solving the problem of color distortion in the sky re-gion,and the details of the defogged image are clear and the color is natural.

关键词

天空分割/暗通道先验/图像去雾/大气光值估计

Key words

sky segmentation/dark channel prior/image defogging/estimation of atmospheric light value

引用本文复制引用

出版年

2024
舰船电子工程
中国船舶重工集团公司第709研究所 中国造船工程学会 电子技术学术委员会

舰船电子工程

CSTPCD
影响因子:0.243
ISSN:1627-9730
参考文献量20
段落导航相关论文