Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising
Open accessThe DOI matches its Crossref record; 53 of the 53 references it cites were found in a registry or an index.
Abstract
The discriminative model learning for image denoising has been recently attracting considerable attentions due to its favorable denoising performance. In this paper, we take one step forward by investigating the construction of feed-forward denoising convolutional neural networks (DnCNNs) to embrace the progress in very deep architecture, learning algorithm, and regularization method into image denoising. Specifically, residual learning and batch normalization are utilized to speed up the training process as well as boost the denoising performance. Different from the existing discriminative denoising models which usually train a specific model for additive white Gaussian noise at a certain noise level, our DnCNN model is able to handle Gaussian denoising with unknown noise level (i.e., blind Gaussian denoising). With the residual learning strategy, DnCNN implicitly removes the latent clean image in the hidden layers. This property motivates us to train a single DnCNN model to tackle with several general image denoising tasks, such as Gaussian denoising, single image super-resolution, and JPEG image deblocking. Our extensive experiments demonstrate that our DnCNN model can not only exhibit high effectiveness in several general image denoising tasks, but also be efficiently implemented by benefiting from GPU computing.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“The architecture of DnCNN is an end-to-end trainable deep CNN for Gaussian denoising and it adopts the residual learning strategy to remove the latent clean image from noisy observation, as reported in [6] by Zhang et al.”
From Median Filter Aided CNN Based Image Denoising: An Ensemble Approach (2021) · cites this paper as [6]
The cited source says what the claim attributes to it.
In the citing paper
“it has been pointed out that the combination of batch normalization and residual learning is particularly helpful for Gaussian denoising since they are beneficial to each other.”
From Learning Deep CNN Denoiser Prior for Image Restoration (2017) · cites this paper as [65]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 7×
“According to [22], when the original mapping is more like an identity mapping, the residual mapping will be much easier to be optimized.”
BackgroundInfluentialVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 4×
“The representative achievements in training CNN models include Rectified Linear Unit (ReLU) [20], tradeoff be-tween depth and width [19], [26], parameter initialization [27], gradient-based optimization algorithms [28], [29], [30], batch normalization [21] and residual learning [22].”
BackgroundMethodInfluentialImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 2×
“The representative achievements in training CNN models include Rectified Linear Unit (ReLU) [20], tradeoff between depth and width [19], [26], parameter initialization [27], gradient-based optimization algorithms [28], [29], [30], batch normalization [21] and residual learning [22].”
BackgroundAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 2×
“The representative achievements in training CNN models include Rectified Linear Unit (ReLU) [20], tradeoff be-tween depth and width [19], [26], parameter initialization [27], gradient-based optimization algorithms [28], [29], [30], batch normalization [21] and residual learning [22].”
BackgroundMethodGoing deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818 · cited 6×
“Batch normalization [21] is proposed to alleviate the internal covariate shift by incorporating a normalization step and a scale and shift step before the nonlinearity in each layer.”
BackgroundMethodInfluentialDelving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.123 · Cited by 19,741
Nonlinear total variation based noise removal algorithms
Rudin, L. I., Osher, S., & Fatemi, E. (1992) · Physica D: Nonlinear Phenomena · doi:10.1016/0167-2789(92)90242-f · Cited by 15,704 · cited 1×
“Over the past few decades, various models have been exploited for modeling image priors, including nonlocal self-similarity (NSS) models [1], [2], [3], [4], sparse models [4], [5], [6], gradient models [7], [8], [9] and Markov random field (MRF) models [10], [11], [12].”
BackgroundBatch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S. (2024) · arXiv (Cornell University) · doi:10.57702/o9raffed · Cited by 14,214
Image denoising by sparse 3-D transform-domain collaborative filtering
Dabov, K., Foi, A., Katkovnik, V., & Egiazarian, K. (2007) · IEEE Transactions on Image Processing · doi:10.1109/tip.2007.901238 · Cited by 9,283 · cited 7×
“, blind Gaussian denoising), DnCNN with a single model can still outperform BM3D [2] and TNRD [16] trained”
BackgroundMethodInfluential
Cited by
Items 501–600 of 1,019
Fracastoro, G., Magli, E., Poggi, G., Scarpa, G., Valsesia, D., & Verdoliva, L. (2021) · IEEE Geoscience and Remote Sensing Magazine · doi:10.1109/mgrs.2021.3070956 · Cited by 135
“For example, for optical image denoising, state-of-the-art, data-driven methods, such as the denoising CNN (DnCNN) [24], and, more recently, nonlocal neural networks [25]–[27], have outperformed, by a large margin, model-based approaches, such as block matching and 3D filtering (BM3D) [28] and its variants.”
BackgroundMethodCan learning from natural image denoising be used for seismic data interpolation?
Zhang, H., Yang, X., & Ma, J. (2020) · Geophysics · doi:10.1190/geo2019-0243.1 · Cited by 134
An investigation of quantitative accuracy for deep learning based denoising in oncological PET
Lu, W., Onofrey, J. A., Lu, Y., Shi, L., Ma, T., Liu, Y., & Liu, C. (2019) · Physics in Medicine and Biology · doi:10.1088/1361-6560/ab3242 · Cited by 134
“A few studies suggested that residual learning could be better than image learning in image post-processing, in terms of preventing performance degradation and improving training efficiency (Han et al 2016, Zhang et al 2017, Xu et al 2017).”
BackgroundRethinking Performance Gains in Image Dehazing NetworksPreprint
Song, Y., Zhou, Y., Qian, H., & Du, X. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2209.11448 · Cited by 134
“Then, we add local residuals [17] to the convolutional blocks and global residual [59] to the network.”
MethodDeep learning based feature envy detection
Liu, H., Xu, Z., & Zou, Y. (2018) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1145/3238147.3238166 · Cited by 133
“First, significant advances in CNN have been achieved recently, which makes CNN effective in increasing the capacity and flexibility of machine learning [62].”
BackgroundIterative Joint Image Demosaicking and Denoising Using a Residual Denoising Network
Kokkinos, F., & Lefkimmiatis, S. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2905991 · Cited by 133
“Furthermore, the approaches P 3 , RED and IRCNN are based upon static denoisers like Non Local Means [40], BM3D [41] and DCNN [32], meanwhile we opt to use a universal denoiser, like ResDNet, that can be further trained on any available training data.”
BackgroundMethodInfluentialC2N: Practical Generative Noise Modeling for Real-World Denoising
Jang, G., Lee, W., Son, S., & Lee, K. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00235 · Cited by 133
“After the DnCNN [50] model has achieved a significant performance gain over traditional methods [9, 14], CNN-based methods have become main-stream in the image denoising area.”
BackgroundMethodInfluentialWatch Me, but Don't Touch Me! Contactless Control Flow Monitoring via Electromagnetic Emanations
Han, Y., Etigowni, S., Liu, H., Zonouz, S., & Petropulu, A. P. (2017) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.1145/3133956.3134081 · Cited by 132
“Common activation functions include sigmoid [26], rectiied linear unit (ReLU) [64], tanh, etc.”
BackgroundNonlocality-Reinforced Convolutional Neural Networks for Image Denoising
Cruz, C., Foi, A., Katkovnik, V. Y., & Egiazarian, K. (2018) · IEEE Signal Processing Letters · doi:10.1109/lsp.2018.2850222 · Cited by 132
“, [10], [11]) entail models trained for specific noise standard deviation values Σ = {σ1 , .”
BackgroundMethodDeep Network Interpolation for Continuous Imagery Effect Transition
Wang, X., Yu, K., Dong, C., Tang, X., & Loy, C. C. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00179 · Cited by 132
Momentum-Net: Fast and Convergent Iterative Neural Network for Inverse Problems
Chun, I. Y., Huang, Z., Lim, H., & Fessler, J. A. (2020) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2020.3012955 · Cited by 133
“For dCNN refiners, we use the following residual multihidden layer CNN architecture, a simplified DnCNN [4]”
BackgroundMethodMulti-exposure image fusion via deep perceptual enhancement
Han, D. Y., Li, L., Guo, X., & Ma, J. (2021) · Information Fusion · doi:10.1016/j.inffus.2021.10.006 · Cited by 132
SRConvNet: A Transformer-Style ConvNet for Lightweight Image Super-Resolution
Li, F., Cong, R., Wu, J., Bai, H., Wang, M., & Zhao, Y. (2024) · International Journal of Computer Vision · doi:10.1007/s11263-024-02147-y · Cited by 133
Learning Filter Basis for Convolutional Neural Network Compression
Li, Y., Gu, S., Gool, L. V., & Timofte, R. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00572 · Cited by 130
“Recently, deep convolutional neural network (CNN) based approaches have been setting new state-of-the-art results not only for high-level computer vision tasks such as image classification [25, 44, 15, 18], segmentation [11, 35], and object detection [11, 10, 43, 42], but also for low-level tasks such as image super-r…”
BackgroundMajumdar, A. (2018) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2838679 · Cited by 129
“We also compare with the latest deep CNN (DnCNN) denoising method [34].”
ResultDilated Residual Networks with Symmetric Skip Connection for image denoising
Peng, Y., Zhang, L., Liu, S., Wu, X., Zhang, Y., & Wang, X. (2019) · Neurocomputing · doi:10.1016/j.neucom.2018.12.075 · Cited by 129
Self-citationLow-Light Image Enhancement via the Absorption Light Scattering Model
Wang, Y.-F., Liu, H., & Fu, Z. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2922106 · Cited by 129
InSituNet: Deep Image Synthesis for Parameter Space Exploration of Ensemble Simulations
He, W., Wang, J., Guo, H., Wang, K.-C., Shen, H., Raj, M., Nashed, Y. S. G., & Peterka, T. (2019) · IEEE Transactions on Visualization and Computer Graphics · doi:10.1109/tvcg.2019.2934312 · Cited by 130
“Our work is related to deep learning based image synthesis, which has been used in various applications, including super-resolution [18, 32, 36], denoising [68, 70], inpainting [46, 68], texture synthesis [24, 71], text-to-image synthesis [50], style transfer [23,32,72], and rendering [6,19].”
BackgroundDeep Architectures for Image Compression: A Critical Review
Mishra, D., Singh, S. K., & Singh, R. K. (2021) · Signal Processing · doi:10.1016/j.sigpro.2021.108346 · Cited by 130
Perception-Driven Deep Underwater Image Enhancement Without Paired Supervision
Jiang, Q., Kang, Y., Wang, Z., Ren, W., & Li, C. (2023) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2023.3327613 · Cited by 129
“Accordingly, the output z s and z t of the encoder h are defined as Inspired by the DnCNN [44], we we adopt the residual learning formulation to train a residual mapping since the underwater image are often absent of some color channel information when imaging due to the light attenuation.”
MethodDecomposed Diffusion Sampler for Accelerating Large-Scale Inverse ProblemsPreprint
Chung, H., Lee, S., & Ye, J. C. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2303.05754 · Cited by 129
Real-time Image Enhancer via Learnable Spatial-aware 3D Lookup Tables
Wang, T., Li, Y., Peng, J., Ma, Y., Wang, X. H., Song, F., & Yan, Y. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00247 · Cited by 128
“Recently, many deep learning-based approaches have been proposed and achieved SOTA results [9, 15, 25, 4, 20, 14, 26, 19, 28] in the field of computational imaging.”
BackgroundPerceive-IR: Learning to Perceive Degradation Better for All-in-One Image Restoration
Zhang, X., Ma, J., Wang, G., Zhang, Q., Zhang, H., & Zhang, L. (2025) · IEEE Transactions on Image Processing · doi:10.1109/tip.2025.3566300 · Cited by 128
“This targeted approach has yielded significant success across a spectrum of restoration tasks, e.g. , denoising [1]–[3], dehazing [4]–[6], deraining [7]–[9], deblurring [10]– [12], and low-light enhancement [13]–[15].”
InfluentialSelf-citationDilated Deep Residual Network for Image Denoising
Wang, T., Sun, M., & Hu, K. (2017) · 2017 IEEE 29th International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2017.00192 · Cited by 101
“[12] explored the feasibility of connecting residual learning and image denoising, and they trained CNN based denoisers as priors in [11].”
BackgroundMethodInfluentialMeng, A., Chen, S., Ou, Z., Ding, W., Zhou, H., Fan, J., & Yin, H. (2021) · Energy · doi:10.1016/j.energy.2021.121795 · Cited by 127
Speed scaling in multiphoton fluorescence microscopy
Wu, J., Ji, N., & Tsia, K. K. (2021) · Nature Photonics · doi:10.1038/s41566-021-00881-0 · Cited by 127
Solving Inverse Problems with Latent Diffusion Models via Hard Data Consistency
Song, B., Kwon, S. M., Zhang, Z., Hu, X., Qu, Q., & Shen, L. (2023) · International Conference on Learning Representations · doi:10.48550/arxiv.2307.08123 · Cited by 127
Deep Boosting for Image Denoising
Chen, C., Xiong, Z., Tian, X., & Wu, F. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01252-6_1 · Cited by 126
“Three representative models: AR-CNN [30], TNRD [11], and DnCNN [12] are adopted for comparison.”
BackgroundMethodInfluentialResidual-driven Fuzzy C-Means Clustering for Image Segmentation
Wang, C., Witold Pedrycz, C., Li, Z., & Zhou, M. (2021) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2020.1003420 · Cited by 126
“Compared with commonly used supervised methods such as convolutional neural networks (CNNs) [35], [36], [37], [38], [39], [40] and dictionary learning [41], [42], it realizes the residual estimation precisely by virtue of a fidelity term rather than using any image samples to train a residual estimation model.”
MethodKhmag, A. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-13569-6 · Cited by 126
PaDNet: Pan-Density Crowd Counting
Tian, Y., Lei, Y., Zhang, J., & Wang, J. Z. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2952083 · Cited by 97
“Inspired by the success of convolutional neural network (CNN) [6, 7, 8], researchers employed CNN-based methods to predict a density map which includes important spatial distribution information for dense crowd counting.”
MethodTuning-free Plug-and-Play Proximal Algorithm for Inverse Imaging ProblemsPreprint
Wei, K., Avilés-Rivero, A. I., Liang, J., Fu, Y., Schönlieb, C., & Huang, H. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2002.09611 · Cited by 125
“Unlike existing semi-automated parameter tuning criteria (Wang & Chan, 2017; Chan et al., 2017; Zhang et al., 2017b; Eksioglu, 2016; Tirer & Giryes, 2018), our method is fully automatic and is purely learned from the data, which significantly eases the burden of manual parameter tuning.”
BackgroundMethodInfluentialKBNet: Kernel Basis Network for Image RestorationPreprint
Zhang, Y., Li, D., Shi, X., He, D., Song, K., Wang, X., Qin, H., & Li, H. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2303.02881 · Cited by 127
“CBSD68 [46] Kodak24 [20] McMaster [78] Urban100 [29] Method σ=15 σ=25 σ=50 σ=15 σ=25 σ=50 σ=15 σ=25 σ=50 σ=15 σ=25 σ=50 MACs IRCNN [76] 33.86 31.16 27.86 34.69 32.18 28.93 34.58 32.18 28.91 33.78 31.20 27.70 - FFDNet [77] 33.87 31.21 27.96 34.63 32.13 28.98 34.66 32.35 29.18 33.83 31.40 28.05 - DnCNN [75] 33.90 31.24…”
BackgroundMethodInfluentialMachine Learning in PET: From Photon Detection to Quantitative Image Reconstruction
Gong, K., Berg, E., Cherry, S. R., & Qi, J. (2019) · Proceedings of the IEEE · doi:10.1109/jproc.2019.2936809 · Cited by 124
“In recent years, deep learning methods have shown great potential for low-level image processing tasks, such as superresolution [128], [129] and denoising applications [130].”
BackgroundMR image reconstruction using deep learning: evaluation of network structure and loss functions
Ghodrati, V., Shao, J., Bydder, M., Zhou, Z., Yin, W., Nguyen, K., Yang, Y., & Hu, P. (2019) · Quantitative Imaging in Medicine and Surgery · doi:10.21037/qims.2019.08.10 · Cited by 124
Rethinking Noise Synthesis and Modeling in Raw Denoising
Zhang, Y., Qin, H., Wang, X., & Li, H. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00455 · Cited by 124
“The Gaussian noise model is most widely used when we evaluate the denoising methods [31, 24, 10, 22, 6].”
MethodDeep Unfolding for Snapshot Compressive Imaging
Meng, Z., Yuan, X., & Jalali, S. (2023) · International Journal of Computer Vision · doi:10.1007/s11263-023-01844-4 · Cited by 124
“In our experiments, we implemented GAP-net with four different denoisers, including a 14-layer auto-encoder (Vincent et al., 2008), a 17-layer DnCNN (Zhang et al., 2017), a 10-layer ResNet (He et al., 2016) and a 15-layer U-net (Ron-neberger et al., 2015).”
MethodInfluentialPark, D., Lee, B. H., & Chun, S. Y. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.00563 · Cited by 124
“Recently, deep learning-based approaches have demonstrated remarkable performance with fast computations, but a majority of them have been focusing on known single degradation such as denoising [2, 4, 20, 25, 65, 67, 68], deraining [5, 16, 30, 51, 52, 59, 60], deblurring [19, 23, 24, 36, 38, 43, 46, 55], desnowing [11…”
BackgroundLearning Graph-Convolutional Representations for Point Cloud Denoising
Pistilli, F., Fracastoro, G., Valsesia, D., & Magli, E. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58565-5_7 · Cited by 123
Underwater Image Enhancement With Lightweight Cascaded Network
Jiang, N., Chen, W., Lin, Y., Zhao, T., & Lin, C. (2022) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2021.3115442 · Cited by 123
“1, where the denoising method DnCNN [10] generates color cast artifacts while the deraining method DID-MDN [4] generates over-saturation.”
MethodA fast blind zero-shot denoiser
Lequyer, J., Philip, R., Sharma, A., Hsu, W.-H., & Pelletier, L. (2022) · Nature Machine Intelligence · doi:10.1038/s42256-022-00547-8 · Cited by 123
“However, with the advent of deep learning, such explicit models are avoidable by instead training a neural network to learn how to map noisy images to their clean counterparts, such as in DnCNN [2], or even just by training it to map noisy pairs of images to one another, such as in Noise2Noise [3].”
BackgroundInfluentialImage Restoration by Denoising Diffusion Models with Iteratively Preconditioned Guidance
Garber, T., & Tirer, T. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.02385 · Cited by 124
“Namely, using (1) to generate training pairs { y i , x ⇤ i } and training a DNN to invert the map [8, 18, 34, 43].”
MethodTotal Deep Variation for Linear Inverse Problems
Kobler, E., Effland, A., Kunisch, K. K., & Pock, T. (2020) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr42600.2020.00757 · Cited by 122
A Comparative Study of Image Restoration Networks for General Backbone Network Design
Chen, X., Li, Z., Pu, Y., Liu, Y., Zhou, J., Qiao, Y., & Dong, C. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-73209-6_5 · Cited by 122
Noise2Void - Learning Denoising from Single Noisy ImagesPreprint
Krull, A., Buchholz, T.-O., & Jug, F. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.10980 · Cited by 122
“comparison to DnCNN using the numbers reported in [25].”
BackgroundMethodResultInfluentialBlind Universal Bayesian Image Denoising with Gaussian Noise Level Learning
Helou, M. E., & Susstrunk, S. E. (2019) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · Cited by 122
Suganuma, M., Liu, X., & Okatani, T. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.00925 · Cited by 121
“We compare our method with Input Ou s Ground truth FFDNet Input Ours Ground truth DnCNN two single distortion methods, DnCNN [51] and MemNet [39].”
BackgroundMethodInfluentialNTIRE 2019 Challenge on Real Image Denoising: Methods and Results
Abdelhamed, A., Timofte, R., Brown, M. S., Yu, S., Park, B., Jeong, J., Jung, S., Kim, D.-W., Chung, J. R., Liu, J., et al. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00273 · Cited by 121
“Unlike conventional image processing networks based on deep learning [8, 38, 39], these factorized residual sub-networks are expected to act ensemblelike and learn different kinds of residual data for the final goal [31].”
BackgroundMethodSelf-citationFrom Rank Estimation to Rank Approximation: Rank Residual Constraint for Image Restoration
Zha, Z., Yuan, X., Wen, B., Zhou, J., Zhang, J., & Zhu, C. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2958309 · Cited by 121
“We also compare the proposed RRC with a representative deep learning method: DN-CNN [72].”
SAR Image Despeckling by Noisy Reference-Based Deep Learning Method
Ma, X., Wang, C., Yin, Z., & Wu, P. (2020) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2020.2990978 · Cited by 121
“Deep learning with neural networks, such as the CNN, can provide an effective and implicit way to acquire many noise features and predict the clean images based on the features learned with minimum prior restrictions [20], [21].”
BackgroundSingle Image Deraining Using Bilateral Recurrent Network
Ren, D., Shang, W., Zhu, P., Hu, Q., Meng, D., & Zuo, W. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2994443 · Cited by 121
“By assuming the linear degradation model y = x + r, most existing deep deraining networks follow residual learning in image denoising [13] to learn rain streak layer.”
BackgroundSelf-citationSilva, M. D., & Brown, D. (2023) · Sensors · doi:10.3390/s23208531 · Cited by 121
“The fundamental components of CNNs are convolutional layers, pooling layers, and fully connected layers [32].”
BackgroundViDeNN: Deep Blind Video Denoising
Claus, M., & van Gemert, J. C. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00235 · Cited by 120
Deep plug-and-play prior for low-rank tensor completion
Zhao, X.-L., Xu, W., Jiang, T.-X., Wang, Y., & Ng, M. K. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.03.018 · Cited by 120
“Recently, deep learning-based methods have been developed to learn data-driven priors from a large number of natural images [38–40] and have shown promising performance on extensive application.”
BackgroundA Comprehensive Benchmark for Single Image Compression Artifact Reduction
Liu, J., Liu, D., Yang, W., Xia, S., Zhang, X., & Dai, Y. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.3007828 · Cited by 121
“Successive works fall into two main streams: better network architectures [28], [30], [32] and better utilization of DCT domain information [9], [33].”
BackgroundMethodExploiting Deep Learning in Limited-Fronthaul Cell-Free Massive MIMO Uplink
Bashar, M., Akbari, A., Cumanan, K., Ngo, H. Q., Burr, A. G., Xiao, P., Debbah, M. A., & Kittler, J. (2020) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2020.3000812 · Cited by 120
COLA-Net: Collaborative Attention Network for Image Restoration
Mou, C., Zhang, J., Fan, X., Liu, H., & Wang, R. (2021) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2021.3063916 · Cited by 121
“, DnCNN [4] and FFDNet [5], and recent local attention-based method ADNet [13] as well as some competitive non-local denoisers such as BM3D [17], N3Net [32] and NLRN[20].”
BackgroundMethodInfluentialFace the Future—Artificial Intelligence in Oral and Maxillofacial Surgery
Miragall, M. F., Knoedler, S. A., Kauke‐Navarro, M., Saadoun, R., Grabenhorst, A., Grill, F. D., Ritschl, L. M., Fichter, A. M., Safi, A., & Knoedler, L. (2023) · Journal of Clinical Medicine · doi:10.3390/jcm12216843 · Cited by 120
“Convolutional neural network and generative adversarial network architectures have been utilized specifically for noise reduction and image deblocking tasks [30,31].”
MethodImage Restoration: From Sparse and Low-Rank Priors to Deep Priors [Lecture Notes]
Zhang, L., & Zuo, W. (2017) · IEEE Signal Processing Magazine · doi:10.1109/msp.2017.2717489 · Cited by 115
Self-citationImproving Low-Dose CT Image Using Residual Convolutional Network
Yang, W., Zhang, H., Yang, J., Wu, J., Yin, X., Chen, Y., Shu, H., Luo, L., Coatrieux, G., Gui, Z., et al. (2017) · IEEE Access · doi:10.1109/access.2017.2766438 · Cited by 119
“ACKNOWLEDGMENT The authors would like to thank Dr. CynthiaMcCollough (the Mayo Clinic, USA) for providing clinical projection data of Somatom Definition AS+ CT. REFERENCES [1] Y. Zhu, M. Zhao, Y. Zhao, H. Li, and P. Zhang, ‘‘Noise reduction with low dose CT data based on a modified ROF model,’’ Opt.”
MethodInfluentialSelf-citationDeepDemosaicking: Adaptive Image Demosaicking via Multiple Deep Fully Convolutional Networks
Tan, D. S., Chen, W.-Y., & Hua, K. (2018) · IEEE Transactions on Image Processing · doi:10.1109/tip.2018.2803341 · Cited by 119
Thin cloud removal with residual symmetrical concatenation network
Li, W., Li, Y., Chen, D., & Chan, J. C.-W. (2019) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2019.05.003 · Cited by 119
Qiu, B., Huang, Z., Liu, X., Meng, X., You, Y., Liu, G., Yang, K., Maier, A., Ren, Q., & Lu, Y. (2020) · Biomedical Optics Express · doi:10.1364/boe.379551 · Cited by 119
“In this study, a modified DnCNN was employed to denoise OCT images, which is the most well-known denoising deep network architecture and widely used in many denoising tasks [31].”
MethodCNN-DMRI: A Convolutional Neural Network for Denoising of Magnetic Resonance Images
Tripathi, P. C., & Bag, S. (2020) · Pattern Recognition Letters · doi:10.1016/j.patrec.2020.03.036 · Cited by 119
Cheng, Z., Lu, R., Wang, Z., Zhang, H., Chen, B., Meng, Z., & Yuan, X. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58586-0_16 · Cited by 119
DIOR: Deep Iterative Optimization-Based Residual-Learning for Limited-Angle CT Reconstruction
Hu, D., Zhang, Y., Liu, J., Luo, S., & Chen, Y. (2022) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2022.3148110 · Cited by 119
“Residual learning can promote the information flow, reduce training difficulty and speed up the convergence [46], [47].”
BackgroundSelf-citationDeep Hyperspectral Image Fusion Network With Iterative Spatio-Spectral Regularization
Huang, T., Dong, W., Wu, J., Li, L., Li, X., & Shi, G. (2022) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2022.3152700 · Cited by 119
“Inspired by the success of deep convolutional neural networks (DCNN) in image superresolution [26] and image de-noising [27]–[29], deep learning has also been applied to the problem of hyperspectral image fusion [10], [11], [30]–[32].”
BackgroundA Heterogeneous Group CNN for Image Super-Resolution
Tian, C., Zhang, Y., Zuo, W., Lin, C., Zhang, D. D., & Yuan, Y. (2022) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2022.3210433 · Cited by 119
“Specifically, quantitative analysis is used to test SR results containing PSNR, SSIM, run-time of restoring high-quality images, complexities, and perceptual quality of feature similarity index (FSIM) [63] of popular SR techniques, containing Bicubic [43], A+ [44], RFL [45], self-exemplars super-resolution (SelfEx) [3…”
Self-citationNoise2Same: Optimizing A Self-Supervised Bound for Image DenoisingPreprint
Xie, Y., Wang, Z., & Ji, S. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.11971 · Cited by 119
“The quality of deep learning methods for signal reconstruction from noisy images, also known as deep image denoising, has benefited from the advanced neural network architectures such as ResNet [8], U-Net [19] and their variants [29, 16, 26, 31, 25, 14].”
BackgroundNLH: A Blind Pixel-Level Non-Local Method for Real-World Image Denoising
Hou, Y., Xu, J., Liu, M., Liu, G., Liu, L., Zhu, F., & Shao, L. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2980116 · Cited by 119
“DnCNN+ is based on the color version of DnCNN [39] for blind denoising, but fine-tuned with the results of FFDNet+ [40].”
MethodLearning Invariant Representation for Unsupervised Image Restoration
Du, W., Chen, H., & Yang, H. (2020) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr42600.2020.01449 · Cited by 118
Classification and reconstruction of optical quantum states with deep neural networks
Ahmed, S., Muñoz, C. S., Nori, F., & Kockum, A. F. (2021) · Physical Review Research · doi:10.1103/physrevresearch.3.033278 · Cited by 119
“Deep neural networks (DNNs) have shown impressive results in image classification [14, 15], object detection [16], image denoising and inpainting [17–19], deconvolution [20], generating realistic-looking images [21–24], text generation and translation [25, 26], as well as for generating audio [27], video [28], simulat…”
BackgroundMasked Frequency Modeling for Self-Supervised Visual Pre-TrainingPreprint
Xie, J., Li, W., Zhan, X., Liu, Z., Ong, Y.-S., & Loy, C. C. (2022) · arXiv (Cornell University) · doi:10.48550/arxiv.2206.07706 · Cited by 118
“Low-level image processing tasks, such as image super-resolution [20], deblurring [75] and denoising [76], focus on restoring the high-fidelity image from its corrupted input.”
BackgroundD3R-Net: Dynamic Routing Residue Recurrent Network for Video Rain Removal
Liu, J., Yang, W., Yang, S., & Guo, Z. (2018) · IEEE Transactions on Image Processing · doi:10.1109/tip.2018.2869722 · Cited by 117
“The related topics include super-resolution [15], [24], [33], [36], [59], [61], [63], [64], [66], compression artifacts removal [4], [13], [71], denoising [8], [9], [68], low light enhancement [40], [53], [57], image and video compression [25], [26], [49], [60], et al .”
BackgroundMethodHuang, Z., Zhang, Y., Li, Q., Li, X., Zhang, T., Sang, N., & Hong, H. (2020) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2020.2978276 · Cited by 117
“…there are two reference indexes: the peak signal-to-noise ratio (PSNR) is adopted to assess the ability of noise reduction, while the structural similarity index (SSIM) is exploited to assess the capability of structure preservation [34], used to objectively evaluate the state-of-the-art methods.”
BackgroundMethodInfluentialEnd-to-End Learning for Joint Image Demosaicing, Denoising and Super-Resolution
Xing, W., & Egiazarian, K. (2021) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr46437.2021.00351 · Cited by 118
“With the recent advancement of deep convolutional neural networks (CNNs) and their application in image restoration, several deep learning-based methods achieve the state-of-the-art (SOTA) performance [19, 42, 37, 47].”
BackgroundMethodInfluentialImage Compressed Sensing Using Non-Local Neural Network
Cui, W., Liu, S., Jiang, F., & Zhao, D. (2021) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2021.3132489 · Cited by 117
“Recently, motivated by the superiority of the CNN-based denoisers [40], the plug-and-play algorithms [41]–[44] regularized”
MethodHigh-speed low-light in vivo two-photon voltage imaging of large neuronal populations
Platiša, J., Ye, X., Ahrens, A. M., Liu, C., Chen, A., Davison, I. G., Tian, L., Pieribone, V. A., & Chen, J. L. (2023) · Nature Methods · doi:10.1038/s41592-023-01820-3 · Cited by 117
Neural Degradation Representation Learning for All-in-One Image Restoration
Yao, M., Xu, R., Guan, Y., Huang, J., & Xiong, Z. (2024) · IEEE Transactions on Image Processing · doi:10.1109/tip.2024.3456583 · Cited by 117
“The conventional image restoration pipeline uses a predefined model to represent degradation, e.g., Gaussian noise [2], [52] or motion blur [53], [54].”
BackgroundMethodInfluentialMulti-Temporal Sentinel-1 and -2 Data Fusion for Optical Image Simulation
He, W. P., & Yokoya, N. (2018) · ISPRS International Journal of Geo-Information · doi:10.3390/ijgi7100389 · Cited by 116
CameraNet: A Two-Stage Framework for Effective Camera ISP Learning
Liang, Z., Cai, J., Cao, Z., & Zhang, L. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3051486 · Cited by 116
“The deep learning techniques have been widely used in lowlevel vision research, including image restoration [7, 8, 9, 24] and enhancement [13, 14, 15, 25].”
BackgroundMethodSelf-citationAdvances in Video Compression System Using Deep Neural Network: A Review and Case Studies
Ding, D., Ma, Z., Chen, D., Chen, Q., Liu, Z. S.-J., & Zhu, F. M. (2021) · Proceedings of the IEEE · doi:10.1109/jproc.2021.3059994 · Cited by 116
“These include de-noising [29], [30], motion deblurring [31], [32], contrast enhancement [33], edge detection [34], [35], etc .”
LG-BPN: Local and Global Blind-Patch Network for Self-Supervised Real-World Denoising
Wang, Z., Fu, Y., Liu, J., & Zhang, Y. (2023) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52729.2023.01741 · Cited by 116
Implicit Dual-Domain Convolutional Network for Robust Color Image Compression Artifact Reduction
Zheng, B., Chen, Y., Tian, X., Zhou, F., & Liu, X. (2019) · IEEE Transactions on Circuits and Systems for Video Technology · doi:10.1109/tcsvt.2019.2931045 · Cited by 115
“[16] proposed a DnCNN to handle image denoising problem, which is also effective to handle SR and compression AR problems.”
BackgroundSelf-citationModulating Image Restoration With Continual Levels via Adaptive Feature Modification Layers
He, J., Dong, C., & Qiao, Y. (2019) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr.2019.01131 · Cited by 116
“Deep learning methods have achieved great success in image restoration tasks, such as denoising, super-resolution, compression artifacts reduction, etc [12, 10, 5, 4, 22].”
MethodInfluentialMAC-Net: Model-Aided Nonlocal Neural Network for Hyperspectral Image Denoising
Xiong, F., Zhou, J. W., Zhao, Q., Lu, J., & Qian, Y. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3131878 · Cited by 115
“Additionally, as in [57], batch normalization and ReLu activation are also adopted in all the layers.”
MethodSpatial-Temporal Transformer for Video Snapshot Compressive Imaging
Wang, L., Cao, M., Zhong, Y., & Yuan, X. (2022) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2022.3225382 · Cited by 116
“These greatly limit the application of ViT to dense prediction tasks such as object detection [38], semantic segmentation [39], [40], [41], and low-level tasks such as image denoising [42] and image restoration [43], [44].”
BackgroundEvaluation of denoising techniques to remove speckle and Gaussian noise from dermoscopy images
Göçeri, E. (2022) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2022.106474 · Cited by 115
A Comprehensive Review of Deep Learning-Based Real-World Image Restoration
Zhai, L., Wang, Y., Cui, S., & Zhou, Y. (2023) · IEEE Access · doi:10.1109/access.2023.3250616 · Cited by 116
“High-resolution single-scale networks [118], [124], [140], [141] produce images with spatially more accurate details, and are less effective in encoding contextual information due to their limited receptive field.”
BackgroundHVI: A New Color Space for Low-light Image EnhancementPreprint
Yan, Q., Feng, Y., Zhang, C., Pang, G., Shi, K., Wu, P., Dong, W., Sun, J., & Zhang, Y. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2502.20272 · Cited by 115
Enhancing Evolutionary Couplings with Deep Convolutional Neural Networks
Liu, Y., Palmedo, P., Ye, Q., Berger, B., & Peng, J. (2018) · Cell Systems · doi:10.1016/j.cels.2017.11.014 · Cited by 114
“In the image recognition field, CNNs have proven very effective at taking a noisy image and returning the clean image (Zhang et al., 2017).”
BackgroundKim, B., Han, M., Shim, H., & Baek, J. (2019) · Medical Physics · doi:10.1002/mp.13713 · Cited by 114
Deformable Kernel Networks for Joint Image FilteringPreprint
Kim, B., Ponce, J., & Ham, B. (2019) · HAL · Cited by 114
Deep Learning Seismic Random Noise Attenuation via Improved Residual Convolutional Neural Network
Yang, L., Chen, W., Wang, H., & Chen, Y. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3053399 · Cited by 113
“In this synthetic data experiment, we use DnCNN proposed by Zhang et al. [42] as a comparison model.”
BackgroundMethodInfluentialSelf-citationAttack-Resilient Image Watermarking Using Stable Diffusion
Zhang, L., Liu, X., Martin, A. V., Bearfield, C., Brun, Y., & Guan, H. (2024) · neural information processing systems · doi:10.52202/079017-1215 · Cited by 113
“Constructive attacks, on the other hand, view the watermark as noise superimposed on the original image, focusing on its removal via image purification techniques such as Gaussian blur [18], BM3D [13], and learning-based methods like DnCNNs [46].”
Self-citationGradient Step Denoiser for convergent Plug-and-PlayPreprint
Hurault, S., Leclaire, A., & Papadakis, N. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.03220 · Cited by 113
“We compare in Table 2 our method (GS-PnP) against the patch-based method EPLL (Zoran & Weiss, 2011; Hurault et al., 2018), the deep PnP methods IRCNN (Zhang et al., 2017b), DPIR (Zhang et al., 2021), MMO (Pesquet et al., 2021), and the “RED-FP” algorithm (Romano et al., 2017) (with TNRD denoiser (Chen & Pock, 2016)) r…”
BackgroundMethodInfluentialUnsupervised Low-Light Image Enhancement Using Bright Channel Prior
Lee, H., Sohn, K., & Min, D. (2020) · IEEE Signal Processing Letters · doi:10.1109/lsp.2020.2965824 · Cited by 112
“Recently, convolutional neural networks (CNN) have been applied to the low-light image enhancement [14]–[19], following the success in the field of image restoration including image denoising [20], super-resolution [21], rain streak removal [22], and haze removal [23].”
BackgroundEmpowering Low-Light Image Enhancer through Customized Learnable Priors
Zheng, N., Zhou, M., Dong, Y., Rui, X., Huang, J., Li, C., & Zhao, F. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.01154 · Cited by 113
“We conduct the experiments on the SIDD [2] dataset with two well-known baselines, DNCNN [58] and MPRNet [56].”
MethodA Review of deep learning methods for denoising of medical low-dose CT images
Zhang, J., Gong, W., Ye, L., Wang, F., Shangguan, Z., & Cheng, Y. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108112 · Cited by 112
Rahim, T., Usman, M. A., & Shin, S. Y. (2020) · Computerized Medical Imaging and Graphics · doi:10.1016/j.compmedimag.2020.101767 · Cited by 111
“The averaged mean squared error between the desired residual mages and estimated ones from the noisy input can be adopted as losses in the DnCNNs function to learn the trainable parameters θ” [125].”
BackgroundMethodInfluential
Metadata
- Authors
- Zhang, K., Zuo, W., Chen, Y., Meng, D., & Zhang, L.
- Affiliations
- Hong Kong Polytechnic University (HK)
- Harbin Institute of Technology (CN)
- Xi'an Jiaotong University (CN)
- Venue
- IEEE Transactions on Image Processing
- Published
- 2017-07
- Volume
- 26
- Issue
- 7
- Pages
- 3142-3155
- ISSN
- 1057-7149 (ISSN-L)
- 1941-0042
- OpenAlex
- W2508457857
- Cited by
- 8,983
- References
- 52
- Citations by year
- 2026: 889 · 2025: 1243 · 2024: 1129 · 2023: 1153 · 2022: 1123 · 2021: 1187 · 2020: 997 · 2019: 761 · 2018: 399 · 2017: 95
- FWCI
- 255.51
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- IEEE Transactions on Image Processing (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- PolyU Institutional Research Archive (Hong Kong Polytechnic University) (submittedVersion)
- TUScholarShare (Temple University) (submittedVersion)
- arXiv (Cornell University)
- TUScholarShare (Temple University)
- Funders
- HK RGC GRF (PolyU 5313/13E)
- National Natural Scientific Foundation of China (61671182, 61471146)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image and Signal Denoising Methods
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image Enhancement Techniques
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image Processing Techniques
- Keywords
- image denoising, residual learning, batch normalization, unknown noise level, single image super-resolution, GPU acceleration
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.