基于雾线和颜色衰减先验的图像去雾方法

doi:10.11959/j.issn.1000-436x.2023013

通信学报 ›› 2023, Vol. 44 ›› Issue (1): 211-222.doi: 10.11959/j.issn.1000-436x.2023013

• 学术通信 • 上一篇

基于雾线和颜色衰减先验的图像去雾方法

廖苗¹, 陆颜¹, 张锦², 赵于前³, 邸拴虎³

¹ 湖南科技大学计算机科学与工程学院，湖南湘潭 411100
² 长沙理工大学计算机与通信工程学院，湖南长沙 410004
³ 中南大学自动化学院，湖南长沙 410083

修回日期:2022-10-20 出版日期:2023-01-25 发布日期:2023-01-01
作者简介:廖苗（1988- ），女，湖南常德人，博士，湖南科技大学副教授、硕士生导师，主要研究方向为图像处理、医学影像分析、人工智能等
陆颜（1998- ），女，安徽亳州人，湖南科技大学硕士生，主要研究方向为图像处理、计算机视觉等
张锦（1979- ），男，河南信阳人，博士，长沙理工大学教授、博士生导师，主要研究方向为智能软件、人工智能、生物医学工程等
赵于前（1973- ），男，湖南娄底人，博士，中南大学教授、博士生导师，主要研究方向为图像处理、人工智能、模式识别等
邸拴虎（1987- ），男，河北石家庄人，中南大学博士生，主要研究方向为图像处理、机器学习等
基金资助:
湖南省自然科学基金资助项目(2021JJ30275);湖南省自然科学基金资助项目(2021JJ30456);湖南省教育厅科研基金资助项目(20B239);工业控制技术国家重点实验室开放课题基金资助项目(ICT2022B60);国防科技重点实验室基金资助项目(2021-KJWPDL-17)

Image dehazing method based on haze-line and color attenuation prior

Miao LIAO¹, Yan LU¹, Jin ZHANG², Yuqian ZHAO³, Shuanhu DI³

¹ School of Computer Science and Engineering, Hunan University of Science and Technology, Xiangtan 411100, China
² School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha 410004, China
³ School of Automation, Central South University, Changsha 410083, China

Revised:2022-10-20 Online:2023-01-25 Published:2023-01-01
Supported by:
The Natural Science Foundation of Hunan Province(2021JJ30275);The Natural Science Foundation of Hunan Province(2021JJ30456);Scientific Research Fund of Hunan Provincial Education Department(20B239);The Open Research Project of the State Key Laboratory of Industrial Control Technology(ICT2022B60);National Defense Key Laboratory(2021-KJWPDL-17)

摘要/Abstract

摘要：

针对现有图像去雾方法易产生的颜色过饱和、细节丢失、伪影等问题，提出了一种基于雾线和颜色衰减先验的去雾方法。首先，利用雾线先验和霍夫投票估计大气光。然后，根据颜色衰减先验建立关于场景深度的非线性模型，获取准确的透射率。最后，通过对大气散射模型进行反向求解去除图像中的雾霾干扰，获得细节丰富的去雾图像。在RESIDE公共数据集上进行了实验，并与多种现有方法进行了比较。实验结果表明，所提方法可有效去除图像中的雾霾干扰，获得清晰自然的去雾图像，且其时间和空间效率均优于其他方法。

关键词: 图像去雾, 雾线, 场景深度, 大气光, 透射率

Abstract:

To solve the problems caused by the existing image dehazing methods, such as color over-saturation, detail loss, and artifacts, a dehazing method was proposed based on haze-line and color attenuation prior.Firstly, haze-line prior and Hough vote were used to estimate the atmospheric light.Then, a nonlinear model on scene depth was constructed according to color attenuation prior theory to achieve transmission accurately.Finally, the haze was removed from the image by inversely solving the atmospheric scattering model, so as to obtain the dehazed image with rich details.The proposed method was tested on public RESIDE dataset in comparison with many existing methods.Experimental results show that the proposed method can effectively remove the haze from image and obtain a clear and natural dehazed image, which has higher time and space efficiency than other methods.

Key words: image dehazing, haze-line, scene depth, atmospheric light, transmission

中图分类号:

TP391

廖苗, 陆颜, 张锦, 赵于前, 邸拴虎. 基于雾线和颜色衰减先验的图像去雾方法[J]. 通信学报, 2023, 44(1): 211-222.

Miao LIAO, Yan LU, Jin ZHANG, Yuqian ZHAO, Shuanhu DI. Image dehazing method based on haze-line and color attenuation prior[J]. Journal on Communications, 2023, 44(1): 211-222.

图/表 10

图1

图2

图3

图4

图5

图6

表1

表2

图7

表3

参考文献 21

[1]	杨燕, 王志伟 . 基于补偿透射率和自适应雾浓度系数的图像复原算法[J]. 通信学报, 2020,41(1): 66-75.
	YANG Y , WANG Z W . Image restoration algorithm based on compensated transmission and adaptive haze concentration coefficient[J]. Journal on Communications, 2020,41(1): 66-75.
[2]	PARK S , MOON B , KO S ,et al. Low-light image restoration using bright channel prior-based variational Retinex model[J]. EURASIP Journal on Image and Video Processing,2017, 2017:44.
[3]	HAO S J , HAN X , ZHANG Y M ,et al. Low-light enhancement based on an improved simplified Retinex model via fast illumination map refinement[J]. Pattern Analysis and Applications, 2021,24(1): 321-332.
[4]	金仙力, 张威, 刘林峰 . 基于引导滤波和自适应容差的图像去雾算法[J]. 通信学报, 2020,41(5): 27-36.
	JIN X L , ZHANG W , LIU L F . Image defogging algorithm based on guided filtering and adaptive tolerance[J]. Journal on Communications, 2020,41(5): 27-36.
[5]	FATTAL R . Single image dehazing[J]. ACM Transactions on Graphics, 2008,27(3): 1-9.
[6]	HE K M , SUN J , TANG X O . Single image haze removal using dark channel prior[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011,33(12): 2341-2353.
[7]	HE K M , SUN J , TANG X O . Guided image filtering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013,35(6): 1397-1409.
[8]	HUA Z , DING Y , LI J . Image dehazing using near-infrared information based on dark channel prior[J]. Procedia Computer Science, 2021,187: 18-23.
[9]	SU C , BI G L , JIN L X ,et al. Dehazing algorithm based on darkchannel image centroid offset[J]. Acta Optica Sinica, 2019,39(5): 0533001.
[10]	ZHU Q S , MAI J M , SHAO L . A fast single image haze removal algorithm using color attenuation prior[J]. IEEE Transactions on Image Processing, 2015,24(11): 3522-3533.
[11]	杨爱萍, 李晓晓, 张腾飞 ,等. 基于误差回传机制的多尺度去雾网络[J]. 自动化学报, 2021,47(10): 1001-1011.
	YANG A P , LI X X , ZHANG T F ,et al. Multi-scale dehazing network based on error-backward mechanism[J]. Acta Automatica Sinica, 2021,47(10): 1001-1011.
[12]	CAI B L , XU X M , JIA K ,et al. DehazeNet:an end-to-end system for single image haze removal[J]. IEEE Transactions on Image Processing:a Publication of the IEEE Signal Processing Society, 2016,25(11): 5187-5198.
[13]	REN W Q , PAN J S , ZHANG H ,et al. Single image dehazing via multi-scale convolutional neural networks with holistic edges[J]. International Journal of Computer Vision, 2020,128(1): 240-259.
[14]	CHEN D D , HE M M , FAN Q N ,et al. Gated context aggregation network for image dehazing and deraining[C]// Proceedings of IEEE Winter Conference on Applications of Computer Vision. Piscataway:IEEE Press, 2019: 1375-1383.
[15]	DONG Y , LIU Y H , ZHANG H ,et al. FD-GAN:generative adversarial networks with fusion-discriminator for single image dehazing[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020,34(7): 10729-10736.
[16]	LI B Y , REN W Q , FU D P ,et al. Benchmarking single-image dehazing and beyond[J]. IEEE Transactions on Image Processing, 2019,28(1): 492-505.
[17]	韩冉, 曾广淼, 王荣杰 . 基于残差块网络的图像去雨算法[J]. 仪器仪表学报, 2021,41(8): 175-182.
	HAN R , ZENG G M , WANG R J . An image derain algorithm based on the residual block network[J]. Chinese Journal of Scientific Instrument, 2021,41(8): 175-182.
[18]	BERMAN D , TREIBITZ T , AVIDAN S . Single image dehazing using haze-lines[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020,42(3): 720-734.
[19]	LI B Y , PENG X L , WANG Z Y ,et al. AOD-net:all-in-one dehazing network[C]// Proceedings of IEEE International Conference on Computer Vision. Piscataway:IEEE Press, 2017: 4780-4788.
[20]	LIU X H , MA Y R , SHI Z H ,et al. GridDehazeNet:attention-based multi-scale network for image dehazing[C]// Proceedings of IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway:IEEE Press, 2019: 7313-7322.
[21]	QIN X , WANG Z L , BAI Y C ,et al. FFA-net:feature fusion attention network for single image dehazing[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020,34(7): 11908-11915.

方法	IE	PSNR	SSIM
文献[5]方法	7.32±0.18	11.30±1.63	0.72±0.05
文献[6]方法	7.36±0.20	13.91±1.58	0.81±0.04
文献[10]方法	7.38±0.21	14.10±1.16	0.82±0.04
文献[18]方法	7.50±0.20	19.20±1.49	0.82±0.04
本文方法	7.79±0.15	29.58±0.85	0.98±0.01

方法	IE	PSNR	SSIM
文献[12]	7.63±0.17	25.87±2.16	0.94±0.02
文献[19]	7.52±0.19	20.16±1.33	0.91±0.02
文献[20]	7.71±0.18	29.43±2.56	0.98±0.01
文献[21]	7.73±0.20	34.02±1.61	0.99±0.01
本文方法	7.79±0.15	29.58±0.85	0.98±0.01

方法	参数数量/个	浮点计算次数/10⁹次	运行平台	运行时间
方法	参数数量/个	浮点计算次数/10⁹次	运行平台	训练/h	测试/s
文献[5]	＜10	1.02	MATLAB(CPU)	—	23.88
文献[6]	＜10	0.30	MATLAB(CPU)	—	1.69
文献[10]	＜10	0.13	MATLAB(CPU)	1.11	0.76
文献[18]	＜10	0.56	MATLAB(CPU)	—	5.08
文献[12]	8.24×10³	2.76	MATLAB(CPU)	32.37	3.26
文献[19]	1.76×10³	1.22	Pytorch(CPU)	19.82	1.29
文献[20]	0.92×10⁶	51.14	Pytorch(GPU)	20.86	1.87
文献[21]	17.74×10⁶	492	Pytorch(GPU)	136.23	2.36
本文方法	＜10	0.05	MATLAB(CPU)	1.02	0.72

基于雾线和颜色衰减先验的图像去雾方法

Image dehazing method based on haze-line and color attenuation prior

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 21

相关文章 5

Metrics

推荐阅读 0

[1]	刘威, 陈成, 江锐, 卢涛. 四通道无监督学习图像去雾网络[J]. 通信学报, 2022, 43(10): 210-222.
[2]	金仙力,张威,刘林峰. 基于引导滤波和自适应容差的图像去雾算法[J]. 通信学报, 2020, 41(5): 27-36.
[3]	杨燕,王志伟. 基于补偿透射率和自适应雾浓度系数的图像复原算法[J]. 通信学报, 2020, 41(1): 66-75.
[4]	杨燕,陈高科. 基于光补偿和逐像素透射率的图像复原算法[J]. 通信学报, 2017, 38(5): 48-56.
[5]	汤红忠,张小刚,朱玲,陈华,王翔,李骁. 结合最小颜色通道图与传播滤波的单幅图像去雾算法研究[J]. 通信学报, 2017, 38(1): 26-34.