Self-supervised image denoising based on blind-ring network and random recovery mask

doi:10.11772/j.issn.1001-9081.2024091383

Journal of Computer Applications ›› 2025, Vol. 45 ›› Issue (10): 3311-3319.DOI: 10.11772/j.issn.1001-9081.2024091383

• Multimedia computing and computer simulation • Previous Articles

Self-supervised image denoising based on blind-ring network and random recovery mask

Zhenyuan LIANG¹, Songlin JIANG²(), Songhao ZHU¹

^1.College of Automation and College of Artificial Intelligence，Nanjing University of Posts and Telecommunications，Nanjing Jiangsu 210023，China
^2.Urban and Rural Construction School，Minxi Vocational and Technical College，Longyan Fujian 364021，China

Received:2024-09-27 Revised:2025-01-08 Accepted:2025-01-10 Online:2025-03-18 Published:2025-10-10
Contact: Songlin JIANG
About author:LIANG Zhenyuan， born in 2000， M. S. candidate. His research interests include image processing， deep learning.
JIANG Songlin， born in 1972， M. S.， associate professor. His research interests include electric automation， image processing.
ZHU Songhao， born in 1973， Ph. D.， associate professor. His research interests include artificial intelligence， deep learning.
Supported by:
National Natural Science Foundation of China(62001247)

基于盲环网络和随机恢复掩码的自监督图像去噪

梁震远¹, 江松林²(), 朱松豪¹

^1.南京邮电大学自动化学院、人工智能学院，南京 210023
^2.闽西职业技术学院城乡建筑学院，福建龙岩 364021

通讯作者: 江松林
作者简介:梁震远（2000—），男，江苏南京人，硕士研究生，主要研究方向：图像处理、深度学习
江松林（1972—），男，福建龙岩人，副教授，硕士，主要研究方向：电气自动化、图像处理 Email:331319315@qq.coom
朱松豪（1973—），男，江苏如皋人，副教授，博士，主要研究方向：人工智能、深度学习。
基金资助:
国家自然科学基金资助项目(62001247)

Abstract

Abstract:

The existing self-supervised image denoising methods based on blind-spot networks often suffer from severe loss of image information due to limitations in the network structure. To solve this problem， firstly， a self-supervised image denoising method was proposed， which improved the traditional blind-spot network into a Blind-Ring Network （BRN）， so as to further reduce spatial correlation of the noise. Then， to address the issue of image information loss caused by the traditional mask strategies， a Random Recovery Mask （RRM） strategy was proposed， thereby reducing the information loss while enhancing detail information of the denoising results. Finally， a dual constraint loss function was proposed to prevent over-fitting of the model while preserving important information in the image effectively. Experimental results show that compared with the sub-optimal self-supervised image denoising method based on BRN， the proposed method improves the Peak Signal-to-Noise Ratio （PSNR） by 0.17 dB， Structural SIMilarity （SSIM） by 0.007， and reduces the Image Patch Perceptual Similarity （IPPS） by 0.006， on the SIDD validation dataset， verifying its superior denoising performance.

Key words: image denoising, self-supervised learning, Blind-Ring Network (BRN), Random Recovery Mask (RRM), dual constraint loss function

摘要：

现有的基于盲点网络的自监督图像去噪方法常因为网络结构的限制，导致图像信息的严重损失。为解决这一问题，首先，提出一种自监督图像去噪方法，通过将传统的盲点网络改进为盲环网络（BRN），进一步降低噪声的空间相关性；其次，针对传统掩码策略导致图像信息丢失的问题，提出一种随机恢复掩码（RRM）策略，在减少信息损失的同时，增强去噪结果的细节信息；最后，提出一种双约束损失函数，在防止模型过度拟合的同时，有效保留图像的重要信息。实验结果表明，相较于次优的基于BRN的自监督图像去噪方法，所提方法在SIDD验证数据集上的峰值信噪比（PSNR）提高了0.17 dB，结构相似性（SSIM）提高了0.007，图像块感知相似度（IPPS）降低了0.006，验证了所提方法具有优越的去噪性能。

关键词: 图像去噪, 自监督学习, 盲环网络, 随机恢复掩码, 双约束损失函数

CLC Number:

TP391.41

Zhenyuan LIANG, Songlin JIANG, Songhao ZHU. Self-supervised image denoising based on blind-ring network and random recovery mask[J]. Journal of Computer Applications, 2025, 45(10): 3311-3319.

梁震远, 江松林, 朱松豪. 基于盲环网络和随机恢复掩码的自监督图像去噪[J]. 《计算机应用》唯一官方网站, 2025, 45(10): 3311-3319.

Figures/Tables 13

Fig. 1 Structure of proposed model

Fig. 2 Comparison of masking strategies and denoising directions of different networks

Fig. 3 Examples of different mask strategies

Tab. 1 Denoising performance comparison of different methods

类型	方法	SIDD Validation			SIDD Benchmark		DND Benchmark
类型	方法	PSNR/dB	SSIM	IPPS	PSNR/dB	SSIM	PSNR/dB	SSIM
无需预训练	BM3D	25.71	0.576	0.657	25.65	0.685	34.51	0.851
无需预训练	WNNM	26.05	0.592	0.635	25.78	0.809	34.67	0.865
有监督	CBDNet	33.07	0.863	0.288	33.28	0.868	38.05	0.942
	PD	33.96	0.899	0.258	34.23	0.898	38.40	0.945
	DnCNN	37.73	0.943	0.245	37.61	0.941	38.73	0.945
	U-net	38.98	0.954	0.201	38.92	0.953	39.37	0.954
	VDN	39.29	0.956	0.208	39.26	0.955	39.38	0.952
	Restormer	39.93	0.960	0.198	40.02	0.960	39.58	0.955
无监督	BGAN	—	—	—	—	—	35.58	0.922
	CAN	—	—	—	32.48	0.897	—	—
	C2N	35.36	0.932	0.192	35.35	0.937	37.28	0.924
	Uformer	—	—	—	—	—	37.93	0.937
自监督	Noise2Void	27.48	0.664	0.592	27.68	0.668	—	—
	Noise2Self	29.94	0.782	0.556	29.56	0.808	—	—
	NAC	—	—	—	—	—	36.20	0.925
	R2R	—	—	—	34.78	0.898	—	—
	CVF-SID	34.15	0.911	0.423	34.71	0.917	36.50	0.924
	AP-BSN	36.74	0.934	0.281	36.91	0.931	38.09	0.937
	SS-BSN	—	—	—	36.73	0.923	37.72	0.928
	BRN	37.39	0.934	0.176	36.91	0.931	38.09	0.937
	MFAF	—	—	—	37.33	0.929	38.41	0.940
	MBBSN-MCRR	37.51	0.937	0.173	37.63	0.941	38.81	0.945
	本文方法	37.56	0.941	0.170	37.44	0.938	38.76	0.946

Fig. 4 Visualization of denoising effects of different methods on SIDD dataset

Fig. 5 Visualization of denoising effects of different methods on DND dataset

Tab. 2 Complexness comparison of different self-supervised methods

方法	Params/MB	运算量/GFLOPs
NAC	0.6	36
AP-BSN	3.7	29
BRN	15.0	46
MFAF	95.0	29
本文方法	11.0	33

Tab. 3 Ablation experimental results of different blind-ring sizes

盲环尺寸	PSNR/dB
盲环尺寸	SIDD Validation	SIDD Benchmark
1×1	24.24	26.35
3×3	27.23	28.21
5×5	30.91	31.53
7×7	33.67	34.24
9×9	37.56	38.76
11×11	34.21	36.43
13×13	32.68	34.21

Fig. 6 Visualization of denoising results on SIDD dataset with different blind-ring sizes

Tab. 4 Ablation experimental results of pixel-shuffle down-sampling strategy with different step sizes

步长	SIDD Validation		SIDD Benchmark
步长	PSNR/dB	SSIM	PSNR/dB	SSIM
2	31.55	0.897	31.21	0.885
3	34.33	0.921	34.23	0.918
4	37.56	0.941	37.44	0.938
5	35.12	0.928	34.13	0.922
6	33.01	0.903	32.03	0.893

Fig. 7 Visualization of pixel-shuffle down-sampling on SIDD dataset with different step sizes

Tab. 5 Ablation experimental results of random recovery mask strategy

RRM	SIDD Validation		SIDD Benchmark
RRM	PSNR/dB	SSIM	PSNR/dB	SSIM
	35.09	0.923	34.89	0.915
√	37.56	0.941	37.44	0.938

Tab. 6 PSNR and SSIM of different loss functions on SIDD dataset

损失函数	SIDD Validation		SIDD Benchmark
损失函数	PSNR/dB	SSIM	PSNR/dB	SSIM
$L σ y$	34.86	0.918	33.67	0.909
$L ω y$	36.16	0.931	35.94	0.921
$L σ y + L ω y$	37.56	0.941	37.44	0.938

Tab. 6 PSNR and SSIM of different loss functions on SIDD dataset

损失函数	SIDD Validation		SIDD Benchmark
损失函数	PSNR/dB	SSIM	PSNR/dB	SSIM
$L σ y$	34.86	0.918	33.67	0.909
$L ω y$	36.16	0.931	35.94	0.921
$L σ y + L ω y$	37.56	0.941	37.44	0.938

References 43

[1]	KONG X， YANG Q. No-reference image quality assessment for image auto-denoising［J］. International Journal of Computer Vision， 2018， 126（5）： 537-549.
[2]	DABOV K， FOI A， KATKOVNIK V， et al. Image denoising by sparse 3-D transform domain collaborative filtering［J］. IEEE Transactions on Image Processing， 2007， 16（8）： 2080-2095.
[3]	GU S， ZHANG L， ZUO W， et al. Weighted nuclear norm minimization with application to image denoising［C］// Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2014： 2862-2869.
[4]	HOSONO K， ONO S， MIYATA T. Weighted tensor nuclear norm minimization for color image denoising ［C］// Proceedings of the 2016 IEEE Conference on Image Processing. Piscataway： IEEE， 2016：3081-3085.
[5]	LUO E， CHAN S H， NGUYEN T Q. Adaptive image denoising by targeted databases［J］. IEEE Transactions on Image Processing， 2015， 24（7）： 2167-2181.
[6]	ANWAR S， BARNES N. Real image denoising with feature attention［C］// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2019： 3155-3164.
[7]	GUO S， YAN Z， ZHANG K， et al. Toward convolutional blind denoising of real photographs［C］// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2019： 1712-1722.
[8]	ABDELHAMED A， LIN S， BROWN M S. A high-quality denoising dataset for smartphone cameras［C］// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2018： 1692-1700.
[9]	PLÖTZ T， ROTH S. Benchmarking denoising algorithms with real photographs［C］// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2017： 2750-2759.
[10]	WEI K， FU Y， YANG J， et al. A physics-based noise formation model for extreme low-light raw denoising［C］// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2020： 2755-2764.
[11]	ZHANG Y， QIN H， WANG X， et al. Rethinking noise synthesis and modeling in raw denoising［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 4593-4601.
[12]	KIM Y， SOH J W， PARK G Y， et al. Transfer learning from synthetic to real-noise denoising with adaptive instance normalization［C］// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2020： 3479-3489.
[13]	LEHTINEN J， MUNKBERG J， HASSELGREN J， et al. Noise2Noise： learning image restoration without clean data［C］// Proceedings of the 35th International Conference on Machine Learning. New York： JMLR.org， 2018： 2965-2974.
[14]	MORAN N， SCHMIDT D， ZHONG Y， et al. Noisier2Noise： learning to denoise from unpaired noisy data［C］// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2020： 12061-12069.
[15]	KRULL A， BUCHHOLZ T O， JUG F. Noise2Void — learning denoising from single noisy images［C］// Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2019： 2124-2132.
[16]	NESHATAVAR R， YAVARTANOO M， SON S， et al. CVF-SID： cyclic multi-variate function for self-supervised image denoising by disentangling noise from image［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 17562-17570.
[17]	LEE W， SON S， LEE K M. AP-BSN： self-supervised denoising for real-world images via asymmetric PD and blind-spot network［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 17704-17713.
[18]	ZHOU Y， JIAO J， HUANG H， et al. When AWGN-based denoiser meets real noises［C］// Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2020： 13074-13081.
[19]	BUADES A， COLL B， MOREL J M. A non-local algorithm for image denoising［C］// Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition — Volume 2. Piscataway： IEEE， 2005： 60-65.
[20]	ZHANG K， ZUO W， CHEN Y， et al. Beyond a Gaussian denoiser： residual learning of deep CNN for image denoising［J］. IEEE Transactions on Image Processing， 2017， 26（7）： 3142-3155.
[21]	MAO X J， SHEN C， YANG Y B. Image restoration using very deep convolutional encoder-decoder networks with symmetric skip connections［C］// Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2016： 2810-2818.
[22]	TAI Y， YANG J， LIU X， et al. MemNet： a persistent memory network for image restoration［C］// Proceedings of the 2017 IEEE International Conference on Computer Vision. Piscataway： IEEE， 2017： 4549-4557.
[23]	JANG G， LEE W， SON S， et al. C2N： practical generative noise modeling for real-world denoising［C］// Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision. Piscataway： IEEE， 2021： 2330-2339.
[24]	WU X， LIU M， CAO Y， et al. Unpaired learning of deep image denoising［C］// Proceedings of the 2020 European Conference on Computer Vision， LNCS 12349. Cham： Springer， 2020： 352-368.
[25]	KOUSHA S， MALEKY A， BROWN M S， et al. Modeling sRGB camera noise with normalizing flows［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 17442-17450.
[26]	SOLTANAYEV S， CHUN S Y. Training deep learning based denoisers without ground truth data［C］// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2018： 3261-3271.
[27]	BATSON J， ROYER L. Noise2Self： blind denoising by self-supervision［C］// Proceedings of the 36th International Conference on Machine Learning. New York： JMLR.org， 2019： 524-533.
[28]	LAINE S， KARRAS T， LEHTINEN J， et al. High-quality self-supervised deep image denoising［C］// Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2019： 6970-6980.
[29]	PANG T， ZHENG H， QUAN Y， et al. Recorrupted-to-recorrupted： unsupervised deep learning for image denoising［C］// Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2021： 2043-2052.
[30]	XU J， HUANG Y， CHENG M M， et al. Noisy-as-Clean： learning self-supervised denoising from corrupted image［J］. IEEE Transactions on Image Processing， 2020， 29： 9316-9329.
[31]	KIM K， KWON T， YE J C. Noise distribution adaptive self-supervised image denoising using Tweedie distribution and score matching［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 1998-2006.
[32]	ZHANG Z， XU R， LIU M， et al. Self-supervised image restoration with blurry and noisy pairs［C］// Proceedings of the 36th International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2022： 29179-29191.
[33]	BROADDUS C， KRULL A， WEIGERT M， et al. Removing structured noise with self-supervised blind-spot networks［C］// Proceedings of the IEEE 17th International Symposium on Biomedical Imaging. Piscataway： IEEE， 2020： 159-163.
[34]	LI J， ZHANG Z， LIU X， et al. Spatially adaptive self-supervised learning for real-world image denoising［C］// Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2023： 9914-9924.
[35]	RONNEBERGER O， FISCHER P， BROX T. U-net： convolutional networks for biomedical image segmentation［C］// Proceedings of the 2015 International Conference on Medical Image Computing and Computer-Assisted Intervention， LNCS 9351. Cham： Springer， 2015： 234-241.
[36]	YUE Z， YONG H， ZHAO Q， et al. Variational denoising network： toward blind noise modeling and removal［C］// Proceedings of the 33rd International Conference on Neural Information Processing Systems. Red Hook： Curran Associates Inc.， 2019： 1690-1701.
[37]	ZAMIR S W， ARORA A， KHAN S， et al. Restormer： efficient Transformer for high-resolution image restoration［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 5718-5729.
[38]	CHEN J， CHEN J， CHAO H， et al. Image blind denoising with generative adversarial network based noise modeling［C］// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2018： 3155-3164.
[39]	HONG Z， FAN X， JIANG T， et al. End-to-end unpaired image denoising with conditional adversarial networks［C］// Proceedings of the 34th AAAI Conference on Artificial Intelligence. Palo Alto： AAAI Press， 2020： 4140-4149.
[40]	WANG Z， CUN X， BAO J， et al. Uformer： a general U-shaped Transformer for image restoration［C］// Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway： IEEE， 2022： 17662-17672.
[41]	HAN Y J， YU H J. SS-BSN： attentive blind-spot network for self-supervised denoising with nonlocal self-similarity［C］// Proceedings of the 32nd International Joint Conference on Artificial Intelligence. California： ijcai.org， 2023： 800-809.
[42]	TANG H， ZHANG W， ZHU H， et al. Self-supervised real-world image denoising based on multi-scale feature enhancement and attention fusion［J］. IEEE Access， 2024， 12： 49720-49734.
[43]	梁震远，朱松豪. 融合多类替换细化和多分支盲点网络的自监督图像去噪［J］. 小型微型计算机系统， 2025， 46（9）： 2153-2159.
	LIANG Z Y， ZHU S H. Self-supervised denoising method via multi-branch blind-spot network with multiclass replacing refinement［J］. Journal of Chinese Computer Systems， 2025， 46（9）： 2153-2159.

Self-supervised image denoising based on blind-ring network and random recovery mask

基于盲环网络和随机恢复掩码的自监督图像去噪

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 13

References 43

Related Articles 15

Recommended Articles

Metrics

[1]	Chao LIU, Yanhua YU. Knowledge-aware recommendation model combining denoising strategy and multi-view contrastive learning [J]. Journal of Computer Applications, 2025, 45(9): 2827-2837.
[2]	Zonghang WU, Dong ZHANG, Guanyu LI. Multimodal fusion recommendation algorithm based on joint self-supervised learning [J]. Journal of Computer Applications, 2025, 45(6): 1858-1868.
[3]	Junyi ZHU, Leilei CHANG, Xiaobin XU, Zhiyong HAO, Haiyue YU, Jiang JIANG. Self-supervised learning method using minimal prior knowledge [J]. Journal of Computer Applications, 2025, 45(4): 1035-1041.
[4]	Guangju YANG, Tianjian LUO, Kaijun WANG, Siqi YANG. Multi-branch multi-view based contextual contrastive representation learning method for time series [J]. Journal of Computer Applications, 2025, 45(4): 1042-1052.
[5]	Jianfeng YANG, Bin CHEN, Yuxuan LI. Self-supervised point cloud anomaly detection method based on point cloud reconstruction [J]. Journal of Computer Applications, 2025, 45(10): 3302-3310.
[6]	Tingjie TANG, Jiajin HUANG, Jin QIN. Session-based recommendation with graph auxiliary learning [J]. Journal of Computer Applications, 2024, 44(9): 2711-2718.
[7]	Yuwei DING, Hongbo SHI, Jie LI, Min LIANG. Image denoising network based on local and global feature decoupling [J]. Journal of Computer Applications, 2024, 44(8): 2571-2579.
[8]	Jiong WANG, Taotao TANG, Caiyan JIA. PAGCL： positive augmentation graph contrastive learning recommendation method without negative sampling [J]. Journal of Computer Applications, 2024, 44(5): 1485-1492.
[9]	Guijin HAN, Xinyuan ZHANG, Wentao ZHANG, Ya HUANG. Self-supervised image registration algorithm based on multi-feature fusion [J]. Journal of Computer Applications, 2024, 44(5): 1597-1604.
[10]	Rong HUANG, Junjie SONG, Shubo ZHOU, Hao LIU. Image aesthetic quality evaluation method based on self-supervised vision Transformer [J]. Journal of Computer Applications, 2024, 44(4): 1269-1276.
[11]	Yi ZHENG, Cunyi LIAO, Tianqian ZHANG, Ji WANG, Shouyin LIU. Image denoising-based cell-level RSRP estimation method for urban areas [J]. Journal of Computer Applications, 2024, 44(3): 855-862.
[12]	Ruifeng HOU, Pengcheng ZHANG, Liyuan ZHANG, Zhiguo GUI, Yi LIU, Haowen ZHANG, Shubin WANG. Iterative denoising network based on total variation regular term expansion [J]. Journal of Computer Applications, 2024, 44(3): 916-921.
[13]	Pengbo WANG, Wuyang SHAN, Jun LI, Mao TIAN, Deng ZOU, Zhanfeng FAN. Robust splicing forensic algorithm against high-intensity salt-and-pepper noise [J]. Journal of Computer Applications, 2024, 44(10): 3177-3184.
[14]	Yuning ZHANG, Abudukelimu ABULIZI, Tisheng MEI, Chun XU, Maierdana MAIMAITIREYIMU, Halidanmu ABUDUKELIMU, Yutao HOU. Anomaly detection method for skeletal X-ray images based on self-supervised feature extraction [J]. Journal of Computer Applications, 2024, 44(1): 175-181.
[15]	Shengwei MA, Ruizhang HUANG, Lina REN, Chuan LIN. Structured deep text clustering model based on multi-layer semantic fusion [J]. Journal of Computer Applications, 2023, 43(8): 2364-2369.