[1]尹 静,何金玉,罗荣鑫,等.结合天空区域分割和暗通道先验的去雾算法[J].计算机技术与发展,2022,32(08):216-220.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 035]
　YIN Jing,HE Jin-yu,LUO Rong-xin,et al.A Defogging Algorithm Combining Sky Region Segmentation and Dark Channel Prior[J].,2022,32(08):216-220.[doi:10. 3969 / j. issn. 1673-629X. 2022. 08. 035]
点击复制

结合天空区域分割和暗通道先验的去雾算法()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年08期
页码:
216-220
栏目:
应用前沿与综合
出版日期:
2022-08-10

文章信息/Info

Title:
A Defogging Algorithm Combining Sky Region Segmentation and Dark Channel Prior
文章编号:
1673-629X(2022)08-0216-05
作者:
尹 静; 何金玉; 罗荣鑫; 俞文静
广州软件学院,广东 广州 510990
Author(s):
YIN Jing; HE Jin-yu; LUO Rong-xin; YU Wen-jing
Software Engineering Institute of Guangzhou,Guangzhou 510990,China
关键词:
图像去雾; 天空区域分割; 阈值分割; 透射率图; 暗通道先验
Keywords:
image defogging; sky region segmentation; threshold segmentation; transmittance map; dark channel prior
分类号:
TP301. 6;TP391. 9
DOI:
10. 3969 / j. issn. 1673-629X. 2022. 08. 035
摘要:
针对暗通道先验算法处理包含天空区域的雾霾图像不理想以及大气光值误提取的问题,提出了一种结合透射率图的单阈值分割和暗通道先验的单幅图像去雾算法。 首先,将输入的雾霾图像转换为透射率图,基于透射率图的单阈值分割可快速确定分割的阈值,将图像分割为天空区域部分和其他部分,并使用改进的 MSR 算法处理天空区域部分;其次,将经过改进的 MSR 算法处理后的透射率图进行阈值分割,选取 0 或 1 为分割阈值提取出天空区域和非天空区域,解决了传统阈值分割天空区域需要多次观测确定阈值的问题;最后,从天空区域提取大气光值,结合暗通道先验算法得到一幅清晰的复原图像。 实验结果表明,该算法能有效改善大气光值的误提取,准确地实现天空区域分割,在处理轻度及中度雾霾图像时,与传统直方图均衡化算法和暗通道先验算法相比,该算法无论从直观观测还是图像性能评价指标上均更优。
Abstract:
Aiming at the problem that the dark channel prior algorithm is not ideal in processing the haze image containing the sky areaand misextraction of atmospheric light value,we propose a single image defogging algorithm combining single threshold segmentationbased on the transmittance map and the dark channel prior. Firstly,the input haze image is converted into a transmittance map. The singlethreshold segmentation based on the transmittance map can quickly determine the segmentation threshold,and the image is divided intothe sky area and other parts. The improved MSR algorithm is used to process the sky area. Secondly,the transmittance map processed bythe improved MSR algorithm is segmented by threshold,and 0 or 1 is selected as the segmentation threshold to extract the sky area and thenon-sky area,which solves the problem that traditional threshold segmentation requires multiple observations to determine the segmentation threshold when segmenting the sky area. Finally,the atmospheric light value is extracted from the sky area and combined with thedark channel prior algorithm,a clear restored image is obtained. Experimental results show that the proposed algorithm can effectively improve the false extraction of atmospheric light values,and accurately achieve sky region segmentation. When processing light and moderate haze images,the proposed algorithm is more optimized than the traditional histogram equalization algorithm and dark channel prior interms of visual observation or image performance evaluation index.

相似文献/References:

[1]王万国,王滨海,张晶晶,等. 基于直方图规定化的图像去雾算法[J].计算机技术与发展,2014,24(09):241.
　WANG Wan-guo,WANG Bin-hai,ZHANG Jing-jing,et al. Image Haze Removal Algorithm Based on Histogram Specification[J].,2014,24(08):241.
[2]李飞,丁若修,张志佳. 基于曲波变换的图像去雾算法研究[J].计算机技术与发展,2017,27(07):65.
　LI Fei,DING Ruo-xiu,ZHANG Zhi-jia. Research on Image Defogging Algorithm Based on Curvelet Transform[J].,2017,27(08):65.
[3]章星晨,孙刘杰.基于Shearlet变换的Retinex去雾算法[J].计算机技术与发展,2019,29(01):40.[doi:10. 3969 / j. issn. 1673-629X. 2019. 01. 009]
　ZHANG Xing-chen,SUN Liu-jie.Retinex Dehazing Algorithm Based on Shearlet Transform[J].,2019,29(08):40.[doi:10. 3969 / j. issn. 1673-629X. 2019. 01. 009]
[4]章星晨,孙刘杰.基于引导滤波和变差函数的图像去雾算法[J].计算机技术与发展,2019,29(06):23.[doi:10. 3969 / j. issn. 1673-629X. 2019. 06. 005]
　ZHANG Xing-chen,SUN Liu-jie.Image Dehazing Algorithm Based on Guided Filtering and Variation Function[J].,2019,29(08):23.[doi:10. 3969 / j. issn. 1673-629X. 2019. 06. 005]

更新日期/Last Update: 2022-08-10