基于双分支通道先验和Retinex的低照度图像增强算法

doi:10.13229/j.cnki.jdxbgxb.20240148

吉林大学学报(工学版) ›› 2025, Vol. 55 ›› Issue (3): 1028-1036.doi: 10.13229/j.cnki.jdxbgxb.20240148

• 计算机科学与技术 • 上一篇

基于双分支通道先验和Retinex的低照度图像增强算法

李扬^1,^2,³(),李现国^1,³(),苗长云^1,³,徐晟²

^1.天津工业大学电子与信息工程学院，天津 300387
^2.天津中德应用技术大学软件与通信学院，天津 300350
^3.天津市光电检测技术与系统重点实验室，天津 300387

收稿日期:2024-02-05 出版日期:2025-03-01 发布日期:2025-05-20
通讯作者: 李现国 E-mail:lytg1985@126.com;lixianguo@tiangong.edu.cn
作者简介:李扬（1985-），男，讲师，博士研究生.研究方向：机器视觉.E-mail：lytg1985@126.com
基金资助:
国家自然科学基金项目(52271341);天津科技计划项目(22KPXMRC00290);天津市光电检测技术与系统重点实验室开放课题项目(2023LOTDS008)

Low⁃light image enhancement algorithm based on dual branch channel prior and Retinex

Yang LI^1,^2,³(),Xian-guo LI^1,³(),Chang-yun MIAO^1,³,Sheng XU²

^1.School of Electronics and Information Engineering，Tiangong University，Tianjin 300387，China
^2.School of Software and Communications，Tianjin Sino-German University of Applied Sciences，Tianjin 300350，China
^3.Tianjin Key Laboratory of Optoelectronic Detection Technology and Systems，Tianjin 300387，China

Received:2024-02-05 Online:2025-03-01 Published:2025-05-20
Contact: Xian-guo LI E-mail:lytg1985@126.com;lixianguo@tiangong.edu.cn

摘要/Abstract

摘要：

针对现有低照度图像增强算法存在的局部暗光、细节丢失和过增强等问题，提出了一种基于双分支通道先验和Retinex的低照度图像增强算法。在Retinex基础上，提出了双分支亮暗先验特征引导方法，设计了亮通道先验特征模块和暗通道先验特征模块，引导网络抑制反射分量噪声和提升光照分量亮度；设计了像素混合注意力机制模块，分别从通道、空间、像素3个维度对特征进行针对性学习；设计了暗通道折射率估算模块放大图像细节特征；采用混合损失函数，调节亮度、对比度、噪声和色彩约束。在公共数据集上的实验结果表明：与10种先进算法相比，本文算法在提升图像亮度的同时减少了颜色失真和细节丢失，获得了最优的视觉效果和质量指标。

关键词: 低照度图像, 图像增强, 双分支通道先验, Retinex, 像素混合注意力机制

Abstract:

A low illumination image enhancement algorithm based on dual branch channel priors and Retinex is proposed to address the issues of local dimming， detail loss， and over enhancement in existing algorithms. Firstly， on the basis of Retinex， a dual branch bright dark prior feature guidance method is proposed， and a bright channel prior feature module and a dark channel prior feature module are designed to guide the network to suppress reflection component noise and improve lighting component brightness； Secondly， a pixel mixed attention mechanism module is designed to learn targeted features from three dimensions： channel， space， and pixel； Thirdly， a dark channel refractive index estimation module is designed to amplify image detail features. Finally， a mixed loss function is used to adjust brightness， contrast， noise， and color constraints. The experimental results on public datasets show that compared with 10 advanced algorithms， this algorithm improves image brightness while reducing color distortion and detail loss， achieving the best visual effects and quality indicators.

Key words: low light image, image enhancement, dual-branch channel retinexnet, pixel hybrid attention mechanism, channel prior features

中图分类号:

TP391

李扬,李现国,苗长云,徐晟. 基于双分支通道先验和Retinex的低照度图像增强算法[J]. 吉林大学学报(工学版), 2025, 55(3): 1028-1036.

Yang LI,Xian-guo LI,Chang-yun MIAO,Sheng XU. Low⁃light image enhancement algorithm based on dual branch channel prior and Retinex[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(3): 1028-1036.

图/表 13

图1

图2

图3

图4

图5

图6

表1

表2

图7

表3

图8

表4

表5

参考文献 28

1	万方, 雷光波, 徐丽. 基于阶跃滤波器的低照度图像边缘增强算法[J]. 计算机仿真, 2022, 39(5):220-224.
	Wan Fang, Lei Guang-bo, Xu Li. Edge enhancement algorithm of low illumination image based on step filter[J]. Computer Simulation, 2022, 39(5): 220-224.
2	张锦洲, 姬世青, 谭创. 融合卷积神经网络和双边滤波的相贯线焊缝提取算法[J]. 吉林大学学报: 工学版,2024, 54(8): 2313-2318.
	Zhang Jin-zhou, Ji Shi-qing, Tan Chuang. Fusion algorithm of convolution neural network and bilateral filtering for seam extraction[J]. Journal of Jilin University (Engineering and Technology Edition),2024, 54(8): 2313-2318.
3	王欣, 党电太. 基于视觉信息补偿的光照不均图像增强方法[J]. 吉林大学学报: 工学版, 2024, 54(8): 2301-2306.
	Wang Xin, Dang Dian-tai. Image enhancement method of uneven illumination based on visual information compensation[J].Journal of Jilin University (Engineering and Technology Edition),2024, 54(8): 2301-2306.
4	Lin Y H, Lu Y C. Low-light enhancement using a plug-and-play retinex model with shrinkage mapping for illumination estimation[J]. IEEE Transactions on Image Processing, 2022, 31(6): 4897-4908.
5	Ma Q, Wang Y, Zeng T. Retinex based variational framework for low light image enhancement and denoising[J]. IEEE Transactions on Multimedia, 2022, 25(6): 5580-5588.
6	Guo X J, Li Y, Ling H B. LIME: low-light image enhancement via illumination map estimation[J]. IEEE Transactions on Image Processing, 2017,26(2): 982-993.
7	Wang S H, Zheng J, Hu H M, et al. Naturalness preserved enhancement algorithm for non-uniform illumination images[J]. IEEE Transactions on Image Processing, 2013, 22(9): 3538-3548.
8	Tao L, Zhu C, Xiang G Q, et al. LLCNN: a convolutional neural network for low-light image enhancement[C]∥IEEE Visual Communication and Image Processing, St. Petersburg, USA, 2017: No.8305143.
9	Cai J R, Gu S H, Zhang L. Learning a deep single image contrast enhancer from multi-exposure images[J]. IEEE Transactions on Image Processing, 2018,27(4): 2049-2062.
10	Lyu F, Lu F, Wu J, et al. MBLLEN: low-light image/video enhancement using CNNs[C]∥British Machine Vision Conference, Newcastle,UK,2018: 220-232.
11	Zhang Y, Zhang J, Guo X. Kindling the darkness: a practical low-light image enhancer[C]∥ACM International Conference on Multimedia, Nice,France, 2019: 1632-1640.
12	Zhang Y, Guo X, Ma J, et al. Beyond brightening low-light images[J]. International Journal of Computer Vision, 2021, 129(4): 1013-1037.
13	Guo C, Li C, Guo J, et al. Zero-reference deep curve estimation for low-light image enhancement[C]∥IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Seattle, USA, 2020: 1777-1786.
14	Li C, Guo C, Chen C L. Learning to enhance low-light image via zero-reference deep curve estimation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44(8): 4225-4238.
15	Shen L, Yue Z, Feng F, et al. MSR-Net: low-light image enhancement using deep convolutional network[J/OL].[2025-01-20].
16	Wei C, Wang W, Yang W, et al. Deep retinex decomposition for low-light enhancement[C]∥BritishMachine Vision Conference (BMVC), Newcastle, UK, 2018: 481-490.
17	Fan M H, Wang W J, Yang W H,et al. Integrating semantic segmentation and retinex model for low-light image enhancement[C]∥ACM International Conference on Multimedia, Seattle, WA, USA,2020: 2317-2325.
18	Wu W, Weng J, Zhang P, et al. URetinex-Net: retinex-based deep unfolding network for low-light image enhancement[C]∥IEEE Conference on Computer Vision and Pattern Recognition, New Orleans, USA, 2022: 5901-5910.
19	陈广秋, 陈昱存, 李佳悦, 等.基于DNST 和卷积稀疏表示的红外与可见光图像融合[J]. 吉林大学学报:工学版, 2021, 51(3): 996-1010.
	Chen Guang-qiu, Chen Yu-cun, Li Jia-yue,et al. Infrared and visible image fusion based on discrete nonseparable shearlet transform and convolutional sparse representation[J]. Journal of Jilin University (Engineering and Technology Edition), 2021, 51(3): 996-1010.
20	Yan Y Y, Ren W Q, Guo Y F, et al. Image deblurring via extreme channels prior[C]∥IEEE Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 6978-6986.
21	Lu Z, Long B, Yang S. Saturation based iterative approach for single image dehazing[J]. IEEE Signal Processing Letters, 2020, 27: 665-669.
22	Zhao Y L, Wang D H, Wang L O. Convolution accelerator designs using fast algorithms[J].Algorithms,2019,12(5): 112.
23	Ren X, Yang W, Cheng W H, et al. Lr3m: robust low-light enhancement via low rank regularized retinex model[J]. IEEE Transactions on Image Processing, 2020, 29: 5862-5876.
24	Yan W D, Tan R T, Dai D X. Nighttime defogging using high-low frequency decomposition and gray scale color networks[C]∥European Conference on Computer Vision, Glasgow, UK, 2020: 473-488.
25	Jiang Z Q, Li H T, Liu L J, et al. A switched view of Retinex: deep self-regularized low-light image enhancement[J]. Neurocomputing, 2021,454:361-372.
26	Ma K, Zeng K, Wang Z. Perceptual quality assessment for multi-exposure image fusion[J]. IEEE Transactions on Image Processing, 2015, 24(11):3345-3356.
27	Lee C, Lee C, Kim C S. Contrast enhancement based on layered difference representation of 2D histograms[J]. IEEE Transactions on Image Processing, 2013, 22(12): 5372-5384.
28	Fu Z, Yang Y, Tu X, et al. Learning a simple low-light image enhancer from paired low-light instances [C]∥IEEE Computer Vision and Pattern Recognition(CVPR),Vancouver, Canada, 2023: 22252-22261.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

算法	PSNR↑	SSIM↑	CIEDE2000↓	NIQE↓
MSR-Net	14.69	0.49	20.17	5.38
LIME	16.00	0.52	20.39	3.77
MBLLEN	17.95	0.71	21.13	4.31
RetinexNet	17.57	0.64	19.32	4.54
KinD	20.23	0.77	17.39	4.24
KinD++	20.59	0.82	14.22	3.88
Zero-DCE	14.86	0.52	15.23	3.70
Zero-DCE++	16.42	0.60	15.19	3.49
URetinex-Net	21.17	0.81	13.92	4.02
PairLIE	19.51	0.74	14.23	4.33
本文	21.58	0.85	14.19	3.38

算法	参数量/M	计算量/G	运行时间/s
MSR-Net	0.68	318.18	0.168 8
MBLLEN	0.45	301.12	0.188 2
RetinexNet	0.84	587.47	0.321 8
KinD	8.16	567.23	0.386 8
KinD++	8.27	598.28	0.392 0
Zero-DCE	0.80	85.76	0.002 5
Zero-DCE++	0.08	19.06	0.001 5
URetinex-Net	0.36	940.74	0.014 4
PairLIE	0.34	368.27	0.012 3
本文	3.38	438.19	0.218 2

评估指标	baseline	w/o L_e	w/o L_i	w/o L_d	w/o L_c	w/o L_n	w/oL_s
PSNR↑	21.58	20.24	19.76	19.96	20.28	19.92	20.76
SSIM↑	0.858	0.800	0.821	0.831	0.792	0.819	0.838
NIQE↓	3.388	3.483	3.491	3.488	3.497	3.392	3.481

实验序号	BAM	PAM	PSNR↑	SSIM↑	NIQE↓
1	√	√	21.58	0.858	3.388
2	√	×	20.89	0.789	3.532
3	×	√	20.38	0.778	3.558
4	×	×	18.21	0.732	3.688

实验序号	BCPF	DCPF	DT	PSNR↑	SSIM↑	NIQE↓
1	√	√	√	21.58	0.858	3.388
2	√	√	×	20.78	0.833	3.397
3	√	×	×	19.63	0.798	3.482
4	×	√	√	19.89	0.808	3.428

基于双分支通道先验和Retinex的低照度图像增强算法

Low⁃light image enhancement algorithm based on dual branch channel prior and Retinex

RICH HTML

PDF (PC)

摘要/Abstract

引用本文

使用本文

图/表 13

参考文献 28

相关文章 11

Metrics

本文评价

推荐阅读 0

[1]	王欣,党电太. 基于视觉信息补偿的光照不均图像增强方法[J]. 吉林大学学报(工学版), 2024, 54(8): 2301-2306.
[2]	张锦洲,姬世青,谭创. 融合卷积神经网络和双边滤波的相贯线焊缝提取算法[J]. 吉林大学学报(工学版), 2024, 54(8): 2313-2318.
[3]	王德兴,高凯,袁红春,杨钰锐,王越,孔令栋. 基于色彩校正和TransFormer细节锐化的水下图像增强[J]. 吉林大学学报(工学版), 2024, 54(3): 785-796.
[4]	刘铭,杨雨航,邹松霖,肖志成,张永刚. 增强边缘检测图像算法在多书识别中的应用[J]. 吉林大学学报(工学版), 2022, 52(4): 891-896.
[5]	蒋华伟,杨震,张鑫,董前林. 图像去雾算法研究进展[J]. 吉林大学学报(工学版), 2021, 51(4): 1169-1181.
[6]	刘富,刘璐,侯涛,刘云. 基于优化MSR的夜间道路图像增强方法[J]. 吉林大学学报(工学版), 2021, 51(1): 323-330.
[7]	徐岩,孙美双. 基于卷积神经网络的水下图像增强方法[J]. 吉林大学学报(工学版), 2018, 48(6): 1895-1903.
[8]	吴一全, 吴诗婳, 张宇飞. 基于混沌粒子群优化的Contourlet域红外图像自适应增强[J]. 吉林大学学报(工学版), 2014, 44(5): 1466-1473.
[9]	王书民, 张爱武, 王飞, 胡少兴, 孙卫东, 仰继连. 面向高空飞艇平台的视频增强与稳像[J]. 吉林大学学报(工学版), 2014, 44(01): 203-210.
[10]	李骜, 李一兵, 刘丹丹, 杨晓冬. 基于非下采样轮廓波的多曝光工件图像Retinex增强方法[J]. , 2012, (06): 1592-1596.
[11]	孙明超, 张崇, 刘晶红. 多尺度图像增强可见光与红外图像融合 [J]. , 2012, (03): 738-742.