The DOI matches its Crossref record; 74 of the 81 references it cites were found in a registry or an index; 2 of the 26 sources its claims cite were checked for support.
Learning Deep CNN Denoiser Prior for Image Restoration
Open accessAbstract
Model-based optimization methods and discriminative learning methods have been the two dominant strategies for solving various inverse problems in low-level vision. Typically, those two kinds of methods have their respective merits and drawbacks, e.g., model-based optimization methods are flexible for handling different inverse problems but are usually time-consuming with sophisticated priors for the purpose of good performance, in the meanwhile, discriminative learning methods have fast testing speed but their application range is greatly restricted by the specialized task. Recent works have revealed that, with the aid of variable splitting techniques, denoiser prior can be plugged in as a modular part of model-based optimization methods to solve other inverse problems (e.g., deblurring). Such an integration induces considerable advantage when the denoiser is obtained via discriminative learning. However, the study of integration with fast discriminative denoiser prior is still lacking. To this end, this paper aims to train a set of fast and effective CNN (convolutional neural network) denoisers and integrate them into model-based optimization method to solve other inverse problems. Experimental results demonstrate that the learned set of denoisers can not only achieve promising Gaussian denoising results but also can be used as prior to deliver good performance for various low-level vision applications.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
The paper trains a set of fast CNN Gaussian denoisers and uses them as modular priors inside a half-quadratic splitting optimization framework to solve image deblurring and super-resolution. Experiments show competitive denoising accuracy and flexible restoration performance across tasks without task-specific retraining.
Key findings
- The proposed CNN denoiser improves gray-image denoising on BSD68 over BM3D, WNNM, TNRD, and MLP: at σ=15 it reaches 31.63 dB, at σ=25 it reaches 29.15 dB, and at σ=50 it reaches 26.19 dB.
- For color denoising on BSD68, the proposed denoiser consistently beats CBM3D, e.g. 40.36 vs. 40.24 dB at noise level 5, 31.16 vs. 30.71 dB at 25, and 27.86 vs. 27.38 dB at 50.
- The learned denoisers can be inserted into a model-based optimization method to handle deblurring and single-image super-resolution; for example, on the Leaves deblurring case with Gaussian blur σ=1.6 and noise level 2, the proposed method reports 29.78 dB versus 26.95 dB for IDDBM3D and 27.50 dB for NCSR.
- Runtime is competitive: for a 256×256 image with noise level 25, the 24-feature-map version takes 0.10 s on CPU and 0.006 s on GPU, while BM3D takes 0.66 s on CPU and TNRD takes 0.47 s on CPU / 0.010 s on GPU.
Methods
- The authors design a seven-layer CNN denoiser using dilated convolutions with dilation factors 1, 2, 3, 4, 3, 2, 1; middle layers use 64 feature maps, and the effective receptive field is 33×33.
- Training uses 400 BSD images, 400 ImageNet validation images, and 4,744 Waterloo Exploration Database images, cropped into 35×35 patches with N = 256×4,000 patches; Adam is used with residual learning and batch normalization.
- A set of 25 denoisers is trained for gray images and 25 for color images over noise levels from 0 to 50 in steps of 2; these denoisers are integrated into a half-quadratic splitting scheme, where the regularization subproblem becomes a Gaussian denoising step.
Limitations
- The paper states that future work includes reducing the number of CNN denoisers and iterations, extending the framework to inpainting and blind deblurring, and using multiple complementary priors.
- The experimental evaluation is limited to denoising, non-blind deblurring, and super-resolution; inpainting and blind deblurring are not tested.
- The authors note that comparisons with SRCNN and VDSR for non-bicubic super-resolution are unfair because those methods were trained for bicubic degradation, which is evident from the reported setup rather than presented as a limitation of the proposed method.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
2 of 26 cited sources checked · 1 supported · 1 flagged · accuracy not yet measured
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“Note that SRCNN and VDSR are trained with bicubic blur kernel, thus it is unfair to use their models to super-resolve the low-resolution image with Gaussian kernel.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.3. Single Image Super-Resolution
What the sources say
- Disputedcites [21] 21. C. Dong, C. C. Loy, K. He, and X. Tang. Image super-resolution using deep convolutional networks. IEEE transactions on Pattern Analysis and Machine Intelligence, 38(2):295-307, 2016. 2, 7 sourceExperimental — accuracy not yet measured
Section 3.3 (p. 5) says SRCNN training samples are blurred with a Gaussian kernel; bicubic interpolation is a later upscaling step, not the blur kernel. Section 4.4.1 (p. 9) also describes training a separate SRCNN network with a different Gaussian kernel for comparison. The paper does not address VDSR's training. (LLM, conf=0.87)
Checked by a reasoning model — may be wrong.
- Not checked — no full text of the sourcecites [35] 35. J. Kim, J. K. Lee, and K. M. Lee. Accurate image super- resolution using very deep convolutional networks. In IEEE Conference on Computer Vision and Pattern Recognition, pages 1646-1654, 2016. 4, 7 source
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passes- Disputed
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Experimental results demonstrate that the learned set of denoisers can not only achieve promising Gaussian denoising results but also can be used as prior to deliver good performance for various low-level vision applications.”
- Also stated in: § 1. Introduction · boosted
“Extensive experiments on classical IR problems, including deblurring and super-resolution, have demonstrated the merits of integrating flexible model-based optimization methods and fast CNN-based discriminative learning methods.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Abstract
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passes- Also stated in: § 1. Introduction · boosted
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“we empirically find it also can enable fast transfer from one model to another with different noise level.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesThe paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“However, if the size of the training patch is smaller than the receptive field, the performance would decrease.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Strength:hedged· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, the proposed CNN denoiser can have a PSNR gain of about 0.2dB over those three methods.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Load-bearingStrength:hedged· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesThe paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“We empirically find that using training samples with small size can help avoid boundary artifacts.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Strength:plain· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We can see that the proposed denoiser is very competitive in both CPU and GPU implementation.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesThe paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Due to the iterative scheme, it turns out the noise level range of [0, 50] is enough to handle various image restoration problems.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes? Based: unclearThe paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“it can be seen that the proposed denoiser consistently outperforms CBM3D by a large margin.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It can be seen that WNNM, MLP and TNRD can outperform BM3D by about 0.3dB in PSNR.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passes
- It may state two separate claims as one. 13
- Probably not a substantial claim. 10
- The quote only partly states this claim. 1
Every source the paper cites for this claim was checked and supports it.
In the 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.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
What the sources say
- Supportedcites [65] 65. K. Zhang, W. Zuo, Y. Chen, D. Meng, and L. Zhang. Be- yond a Gaussian denoiser: Residual learning of deep CNN for image denoising. IEEE Transactions on Image Process- ing, 2017. 4 source
Section C (pp. 4–5) says residual learning and batch normalization can benefit each other in Gaussian denoising, and reports faster convergence and better denoising performance when combined. Section B (p. 3) also describes their integration as producing fast, stable training and better denoising performance. (LLM, conf=0.95)
Strength:hedged· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“It is worth emphasizing that the proposed denoiser with 24 feature maps of each layer has a comparable PSNR of 28.94dB to TNRD but delivers a faster speed.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“We empirically find using large dataset does not improve the PSNR results of BSD68 dataset [50] but can slightly improve the performance of other testing images.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
What the sources say
- Not checked — no full text of the sourcecites [50] 50. S. Roth and M. J. Black. Fields of experts. International Journal of Computer Vision, 82(2):205-229, 2009. 1, 4, 5 source
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
In the paper
“It turns out that our designed model can have an average P-SNR of 29.15dB on BSD68 dataset [50], which is much better than 28.94dB of 7 layers network with traditional 3×3 convolution filter and very close to 29.20dB of 16 layers network.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.2. The Proposed CNN Denoiser
What the sources say
- Not checked — no full text of the sourcecites [50] 50. S. Roth and M. J. Black. Fields of experts. International Journal of Computer Vision, 82(2):205-229, 2009. 1, 4, 5 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As an example for the testing speed comparison, our method can super-resolve the Butterfly image in 0.5 second on GPU and 12 seconds on CPU, whereas NCSR spends 198 seconds on CPU.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.3. Single Image Super-Resolution
Strength:plain· not yet calibratedSingle caseTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
- Rests on an assumption: The low-resolution image can be modeled by a blurring and subsequent down-sampling operation on a high-resolution one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We can see that IDDBM3D, NCSR and MLP tend to smooth the edges and generate color artifacts.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.2. Image Deblurring
Tests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
- The authors state this as an assumption: The convolution is carried out with circular boundary conditions.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“First, although S-RCNN and VDSR achieve promising results to tackle the case with bicubic kernel, their performance deteriorates seriously when the low-resolution image are not generated by bicubic kernel (see Figure 4).”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.3. Single Image Super-Resolution
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- Rests on an assumption: The low-resolution image can be modeled by a blurring and subsequent down-sampling operation on a high-resolution one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, the proposed method can recover image sharpness and naturalness.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.2. Image Deblurring
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- The authors state this as an assumption: The convolution is carried out with circular boundary conditions.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Third, both of the gray and color CNN denoiser prior based optimization methods can produce promising results.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.3. Single Image Super-Resolution
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
- Rests on an assumption: The low-resolution image can be modeled by a blurring and subsequent down-sampling operation on a high-resolution one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Third, CNN exploits the external prior which is complementary to the internal prior of many existing denoisers such as BM3D.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.1. Why Choose CNN Denoiser?
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
In the paper
“model-based optimization methods are flexible for handling different inverse problems but are usually time-consuming with sophisticated priors for the purpose of good performance”
- Also stated in: § 1. Introduction · plain
“model-based optimization methods are usually timeconsuming with sophisticated priors for the purpose of good performance [27].”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Abstract
What the sources say
- Not checked — no full text of the sourcecites [27] 27. Q. Gao and S. Roth. How well do filter-based MRFs model natural images? In Joint DAGM (German Association for Pattern Recognition) and OAGM Symposium, pages 62-72, 2012. 2 source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- Also stated in: § 1. Introduction · plain
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Second, CNN exhibits powerful prior modeling capacity with deep architecture.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.1. Why Choose CNN Denoiser?
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“First, the inference of CNN is very efficient due to the parallel computation ability of GPU.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.1. Why Choose CNN Denoiser?
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Especially noteworthy is the number of the denoisers which is much less than that of learning different models for different degradations.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✗ Based: failsProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“discriminative learning methods have fast testing speed but their application range is greatly restricted by the specialized task.”
- Also stated in: § 1. Introduction · plain
“discriminative learning methods are usually restricted by specialized tasks.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Abstract
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
- Also stated in: § 1. Introduction · plain
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Such a good compromise between speed and performance over TNRD is properly attributed to the following three reasons.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one. The quote only partly states this claim.
In the paper
“and it is preferable to jointly handle RGB channels.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.1. Why Choose CNN Denoiser?
What the sources say
- Not matchedcites [42] 42. T. Miyata. Inter-channel relation based vectorial total varia- tion for color image recovery. In IEEE International Confer- ence on Image Processing,, pages 2251-2255, 2015. 3
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
- Not matched
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“To make a further comparison with TNRD under similar PSNR performance, we additionally provide the run time of the proposed denoiser where each middle layer has 24 feature maps.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the denoiser here just acts its own role regardless of the noise type and noise level of the image to be denoised.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > Using Batch Normalization and Residual Learning to Accelerate Training.
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“To show the advantage of our design over the above two cases, we have trained three different models on noise level 25 with same training settings.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 3.2. The Proposed CNN Denoiser
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“It is worthy noting that the iterative regularization step of methods such as NCSR and WNNM actually corresponds to solving Eqn.(6a).”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.3. Single Image Super-Resolution
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✗ Based: failsProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The number of iterations is set to 30 as we find it is large enough to obtain a satisfying performance.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.2. Image Deblurring
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✗ Based: failsProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
- The authors state this as an assumption: The convolution is carried out with circular boundary conditions.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“On the other hand, different from discriminative learning methods which are specialized for certain image restoration tasks, the proposed deep CNN denoiser prior based optimization method is flexible in handling various tasks while can produce very favorable results.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 5. Conclusion
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
The paper relies on this without support.
The study of integration with fast discriminative denoiser prior is still lacking.
“However, the study of integration with fast discriminative denoiser prior is still lacking.”
“However, to the best of our knowledge, the study of integration with discriminative denoiser is still lacking.”
The authors state this as an assumption.
The convolution is carried out with circular boundary conditions.
“In addition, we assume the convolution is carried out with circular boundary conditions.”
The paper relies on this without support.
The low-resolution image can be modeled by a blurring and subsequent down-sampling operation on a high-resolution one.
“In general, the low-resolution (LR) image can be modeled by a blurring and subsequent down-sampling operation on a high-resolution one.”
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 1×
“Thus, using 3×3 filter with a large depth is popularized in existing CNN network design [30, 35, 56].”
MethodVery 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 1×
“However, increasing the filter size would not only introduce more parameters but also increase the computational burden [53].”
BackgroundImageNet 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 1×
“Particularly, several CNN techniques, including Rectifier Linear Units (ReLU) [37], batch normalization [32], Adam [36], dilated convolution [63] are adopted into the network design or training.”
MethodAdam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 5×
“Particularly, several CNN techniques, including Rectifier Linear Units (ReLU) [37], batch normalization [32], Adam [36], dilated convolution [63] are adopted into the network design or training.”
MethodInfluentialImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 1×
“Hence, instead of training on a small dataset consisting of 400 Berkeley segmentation dataset (BSD) images of size 180×180 [13], we collect a large dataset which includes 400 BSD images, 400 selected images from validation set of ImageNet database [20] and 4,744 images of Waterloo Exploration Database [40].”
MethodGoing 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 · cited 1×
“Thus, using 3×3 filter with a large depth is popularized in existing CNN network design [30, 35, 56].”
MethodBatch 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 2×
“Particularly, several CNN techniques, including Rectifier Linear Units (ReLU) [37], batch normalization [32], Adam [36], dilated convolution [63] are adopted into the network design or training.”
BackgroundMethodBoyd, S., Parikh, N., Chu, E., Peleato, B., & Eckstein, J. (2011) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000016 · Cited by 17,075
Boyd, S. (2010) · now publishers, Inc. eBooks · doi:10.1561/9781601984616 · Cited by 12,917
A Fast Iterative Shrinkage-Thresholding Algorithm for Linear Inverse Problems
Beck, A., & Teboulle, M. (2009) · SIAM Journal on Imaging Sciences · doi:10.1137/080716542 · Cited by 11,952 · cited 1×
“Note that this property can be also employed in other optimization methods ( e.g ., iterative shrink-age/thresholding algorithms ISTA [4, 14] and FISTA [3]) as long as there involves a denoising subproblem.”
Flagged first, then by citations
6 of the 81 references couldn’t be matched to a known paper.
Research works printed but not linked: 2
- foi, 2006, Pointwise shape adaptive DCT denoising with structure preservation in luminance-chrominance space, International Workshop on Video Processing and Quality Metrics for Consumer Electronics
- danielyan, 2010, Image deblurring by augmented lagrangian with BM3D frame prior, Workshop on Information Theoretic Methods in Science and Engineering, 16
Cited by
Sorted by citations
Items 101–200 of 2,094
Embedding Deep Learning in Inverse Scattering Problems
Sanghvi, Y., Kalepu, Y., & Khankhoje, U. K. (2019) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2019.2915580 · Cited by 164
Learning a Discriminative Prior for Blind Image Deblurring
Li, L., Pan, J., Lai, W., Gao, C., Sang, N., & Yang, M. (2018) · 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.2018.00692 · Cited by 166
“Several approaches train deep CNNs as an image prior or denoiser for non-blind deconvolution [30, 42, 41], which cannot be directly applied in blind deconvolution.”
BackgroundDeep Residual Learning based Enhanced JPEG Compression in the Internet of Things
Qiu, H., Zheng, Q., Memmi, G., Lu, J., Meikang Qiu, A., & Thuraisingham, B. M. (2020) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2020.2994743 · Cited by 155
“One of the most popular issues is to use the DL model to enhance the image quality [18], which can also be seen as to remove the noise of the image [19].”
BackgroundTransCS: A Transformer-Based Hybrid Architecture for Image Compressed Sensing
Shen, M., Gan, H., Ning, C., Hua, Y., & Zhang, T. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3217365 · Cited by 143
“In the application of CS, the design of the sensing matrix and the reconstruction strategy are two core issues [7], [8], [9], [10].”
MethodMoiré photo restoration using multiresolution convolutional neural networks
Sun, Y., Yu, Y., & Wang, W. (2018) · IEEE Transactions on Image Processing · doi:10.1109/tip.2018.2834737 · Cited by 162
“After successes in image classification and recognition [4], [5], they have also been proven highly effective in low-level vision and image processing tasks, including image super-resolution [6], [7], demosaicking [8], denoising [9], and restoration [10].”
BackgroundMethodInfluentialRefusion: Enabling Large-Size Realistic Image Restoration with Latent-Space Diffusion Models
Luo, Z., Gustafsson, F., Zhao, Z., Sjölund, J., & Schön, T. B. (2023) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw59228.2023.00169 · Cited by 175
MambaIRv2: Attentive State Space Restoration
Guo, H., Guo, Y. G., Zha, Y., Zhang, Y., Li, W., Dai, T., Xia, S., & Li, Y. (2025) · 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/cvpr52734.2025.02619 · Cited by 183
Deep Learning for Low-Dose CT Denoising Using Perceptual Loss and Edge Detection Layer
Gholizadeh-Ansari, M., Alirezaie, J., & Babyn, P. (2019) · Journal of Imaging Informatics in Medicine · doi:10.1007/s10278-019-00274-4 · Cited by 150
“fully convolutional, the input size does not have to be xed, test images with their original size are fed to the network. To avoid boundary artifacts, zero padding in convolutional operators are used [42]. The activation function is rectied linear unit and the number of lters in all convolutional layers is 64 except l…”
BackgroundMethodResultInfluentialWINNet: Wavelet-Inspired Invertible Network for Image Denoising
Huang, J., & Dragotti, P. L. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3184845 · Cited by 147
Unified Dynamic Convolutional Network for Super-Resolution With Variational Degradations
Xu, Y.-S., Tseng, S.-Y. R., Tseng, Y., Kuo, H.-K., & Tsai, Y. (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/cvpr42600.2020.01251 · Cited by 149
“For IRCNN, results are extracted from the publication [40].”
MethodZheng, Y., Huang, T., Zhao, X.-L., Chen, Y., & He, W. P. (2020) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2020.2987954 · Cited by 148
“In the future, we will attempt to combine the proposed LRTF-DFR with the convolutional neural network [45], [64]– [66] to learn a more appropriate regularization on factors and further improve the ability of mixed noise removal.”
MethodNTIRE 2018 Challenge on Image Dehazing: Methods and Results
Ancuți, C., Zhao, R., Ma, X., Qin, Y., Jia, L., Friedel, K., Ki, S., Sim, H., Choi, J., Kim, S., et al. (2018) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2018.00134 · Cited by 138
“These output images are resized to the original size, and in the final step the proposed method uses IRCNN to denoise the result.”
MethodSelf-citationUnfolding the Alternating Optimization for Blind Super ResolutionPreprint
Luo, Z., Huang, Y., Shang, L., Wang, L., & Tan, T. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2010.02631 · Cited by 246
“In literature, we can apply a denoise algorithm [36, 7, 15] in the first place.”
MethodDiscrete Cosine Transform Network for Guided Depth Map Super-Resolution
Zhao, Z., Zhang, J., Xu, S., Lin, Z., & Pfister, H. (2022) · 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/cvpr52688.2022.00561 · Cited by 152
“Representative works in this category use sparse dictionary learning [15, 19, 51], which learn dictionaries in a group learning manner and set constraints on the sparse representations of different modalities [2, 63].”
BackgroundMethodPerson Reidentification via Structural Deep Metric Learning
Yang, X., Zhou, P., & Wang, M. (2018) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2018.2861991 · Cited by 140
Pyramid Attention Network for Image Restoration
Mei, Y., Fan, Y., Zhang, Y., Yu, J., Zhou, Y., Liu, D., Fu, Y., Huang, T. S., & Shi, H. (2023) · International Journal of Computer Vision · doi:10.1007/s11263-023-01843-5 · Cited by 231
“In IRCNN Zhang et al. (2017b), a learned set of CNNs are used as denoising prior for other image restoration tasks.”
BackgroundMethodResultInfluentialCRRN: Multi-scale Guided Concurrent Reflection Removal Network
Wan, R., Shi, B., Duan, L., Tan, A., & Kot, A. C. (2018) · 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.2018.00502 · Cited by 137
“Prior works that use deep learning to solve the inverse imaging problems [33, 5] or layer separation problems [32] mainly optimize the whole network on patches with resolution n × n cropped from the whole images.”
BackgroundSampling Methods for Efficient Training of Graph Convolutional Networks: A Survey
Liu, X., Yan, M., Deng, L., Li, G., Ye, X., & Fan, D. (2021) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2021.1004311 · Cited by 146
Metal artifact reduction on cervical CT images by deep residual learning
Huang, X., Wang, J., Tang, F., et al. (2018) · BioMedical Engineering OnLine · doi:10.1186/s12938-018-0609-y · Cited by 135
Learning with Privileged Information for Efficient Image Super-Resolution
Lee, W.-K., Lee, J., Kim, D., & Ham, B. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-58586-0_28 · Cited by 146
“After that, CNN-based SR methods focus on designing wider [34,49,62] or deeper [20,29,32,39,60,61] network architectures for the performance gains.”
MethodDeep Image Demosaicking Using a Cascade of Convolutional Residual Denoising Networks
Kokkinos, F., & Lefkimmiatis, S. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01264-9_19 · Cited by 131
“Finally, while our demosaick network shares a similar philosophy with methods such as RED [25], P 3 [26] and IRCNN [29], it exhibits some important and distinct differences.”
BackgroundInfluentialWaveDM: Wavelet-Based Diffusion Models for Image Restoration
Huang, Y., Huang, J., Liu, J., Yan, M., Yu, D., Lv, J., Chen, C., & Chen, S. (2024) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2024.3359769 · Cited by 162
MPRNet: Multi-Path Residual Network for Lightweight Image Super Resolution
Mehri, A., Behjati, P., & Sappa, A. D. (2021) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv48630.2021.00275 · Cited by 134
Densely Connected Hierarchical Network for Image Denoising
Park, B., Yu, S., & Jeong, J. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00263 · Cited by 134
“In the following year, Zhang et al. proposed CNN denoiser for image restoration (IRCNN) [16]; it reduces the computational complexity with minimizing the performance degradation.”
BackgroundMethodInfluentialAdversarial Representation Learning for Robust Patient-Independent Epileptic Seizure Detection
Zhang, X., Yao, L., Dong, M., Liu, Z., Zhang, Y., & Li, Y. (2020) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2020.2971610 · Cited by 127
“Meanwhile, the excellent latent feature learning ability of Convolutional Neural Networks (CNNs) has been widely used in research areas such as computer vision [23] and natural language process [24].”
BackgroundBlind Universal Bayesian Image Denoising With Gaussian Noise Level Learning
Helou, M. E., & Susstrunk, S. E. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2976814 · Cited by 132
“The recent method [50] that generalizes to image restoration tasks is a non-universal non-blind denoiser, where 25 denoising networks are used for noise levels below 50, and even training parameters are chosen based on the noise level.”
MethodWhen AWGN-Based Denoiser Meets Real Noises
Zhou, Y., Jiao, J., Huang, H., Wang, Y., Wang, J., Shi, H., & Huang, T. S. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i07.7009 · Cited by 139
“Besides, multichannel (Xu et al. 2017; Shi Guo 2018) and multi-scale (Lebrun, Colom, and Morel 2015a; Yu and Koltun 2015) strategy were also investigated for adaptation.”
MethodSpectral Response Function-Guided Deep Optimization-Driven Network for Spectral Super-Resolution
He, J., Li, J., Yuan, Q., et al. (2021) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2021.3056181 · Cited by 132
“As described in [33], the HH -subproblem in (6) can be rewritten as HH�kk+1 = aaaaaa mmmmmm XX 1”
BackgroundMethodSelf-citationImage Correction via Deep Reciprocating HDR Transformation
Yang, X. S., Xu, K., Song, Y., Zhang, Q., Wei, X., & Lau, R. W. H. (2018) · 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.2018.00193 · Cited by 125
“Image restoration methods improve the image quality mainly by reducing the noise via different deep network designs [19, 40, 52], low-rank sparse representation learning [21] or soft-rounding regularization [26].”
BackgroundDNU: Deep Non-Local Unrolling for Computational Spectral Imaging
Wang, L., Sun, C., Zhang, M., Fu, Y., & Huang, 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/cvpr42600.2020.00173 · Cited by 137
“The recently proposed deep denoising prior [61] is also a local prior, as it gradually processes the information from a local neighborhood [54].”
BackgroundAsymmetric CNN for Image Superresolution
Tian, C., Xu, Y. D., Zuo, W., Lin, C., & Zhang, D. D. (2021) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2021.3069265 · Cited by 128
“grated into model-based optimization was a good tool to efficiently recover texture features of the high-definition image [23].”
MethodSelf-citationDeep Convolutional Dictionary Learning for Image Denoising
Zheng, H., Yong, H., & Lei Zhang, A. (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/cvpr46437.2021.00069 · Cited by 129
“, DnCNN [67], N3Net [40], NLRN [33], RNAN [72], FOCNet [26], IRCNN [68], FFDNet [69]), and deep unfolding methods (DKSVD [47], CSCNet [50]).”
MethodInfluentialSelf-citationCSI-Fingerprinting Indoor Localization via Attention-Augmented Residual Convolutional Neural Network
Zhang, B., Sifaou, H., & Li, G. Y. (2023) · IEEE Transactions on Wireless Communications · doi:10.1109/twc.2023.3235449 · Cited by 138
“noise of standard deviation √ λ with a Gaussian Denoiser, where the trajectory prior φ(x) is implicitly replaced by a denoiser prior [27], [41].”
MethodInfluentialDeep Chain HDRI: Reconstructing a High Dynamic Range Image from a Single Low Dynamic Range Image
Lee, S., An, G. H., & Kang, S. (2018) · IEEE Access · doi:10.1109/access.2018.2868246 · Cited by 121
“As a function approximator that infers the unknown mapping between input and output image sets, deep neural networks have advanced performance in the image restoration field such as super resolution [1], [4], deblurring, and denoising [5].”
BackgroundDynamic Path-Controllable Deep Unfolding Network for Compressive Sensing
Song, J., Chen, B., & Zhang, J. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2023.3263100 · Cited by 133
“DUNs on CS usually integrate some effective convolutional neural network (CNN) denoisers into some optimization methods, including half quadratic splitting (HQS) algorithm [38], [39], proximal gradient descent (PGD) algorithm [28], [31], [40], [41], [42], and inertial proximal algorithm for nonconvex optimization (iPi…”
BackgroundBlind Universal Bayesian Image Denoising with Gaussian Noise Level Learning
Helou, M. E., & Susstrunk, S. E. (2019) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · Cited by 121
Nonlocality-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 121
“Advantages of CNNFs: 1) ability to learn and extract complex image features; 2) efficient implementation on graphics processing units (GPUs) [5].”
BackgroundDeep unfolding of a proximal interior point method for image restoration
Bertocchi, C., Chouzenoux, É., Corbineau, M.-C., Pesquet, J.-C., & Prato, M. (2019) · Inverse Problems · doi:10.1088/1361-6420/ab460a · Cited by 120
“, but it gives an upperbound on the image quality that one can expect by solving (79). We also use the following deep learning image restoration methods for comparison: EPLL [58], MLP [10], and IRCNN [54]. In order to provide a fair comparison, these methods are tested under the same conditions as they were trained fo…”
BackgroundMethodInfluentialA General Decoupled Learning Framework for Parameterized Image Operators
Fan, Q., Chen, D., Yuan, L., Hua, G., Yu, N., & Chen, B. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2925793 · Cited by 124
“TABLE 9 Quantitative Results (PSNR) of the Image Denoising Task on the BSD68 Benchmark s BM3D [42] WNNM [43] TNRD [37] DCDP [44] Ours”
BackgroundMultiple Cycle-in-Cycle Generative Adversarial Networks for Unsupervised Image Super-Resolution
Zhang, Y., Liu, S., Dong, C., Zhang, X., & Yuan, Y. (2019) · IEEE Transactions on Image Processing · doi:10.1109/tip.2019.2938347 · Cited by 117
A Benchmark for Sparse Coding: When Group Sparsity Meets Rank Minimization
Zha, Z., Yuan, X., Wen, B., Zhou, J., Zhang, J., & Zhu, C. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.2972109 · Cited by 119
“The BPFA, IPPO, JSM, Aloha, [47], BPFA [48], IPPO [49], JSM [50], A LOHA [51], BKSVD [68], NGS [69], IRCNN [71], IDBP [72] AND GSC-w(cid:96) p FOR IMAGE INPAINTING .”
MethodInfluentialImage Demoireing with Learnable Bandpass Filters
Zheng, B., Yuan, S., Slabaugh, G., & Leonardis, A. (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/cvpr42600.2020.00369 · Cited by 130
“Subsequent research [32, 19, 45] further improved these models by increasing the network depth, introducing skip connections [26] and residual learning.”
BackgroundPre-Trained Image Processing TransformerPreprint
Chen, H., Wang, Y., Guo, T., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.00364 · Cited by 120
“…of the proposed method, Urban100( × 4): img 004 HR VDSR[41] EDSR[51] RDN[94] OISR[35] SAN[17] RNAN [93] IGNN BSD68: 163085 GT Noisy ( σ =50) CBM3D [16] TNRD [14] RDN [94] DnCNN [87] MemNet [65] IRCNN [88] FFDNet [89] IPT (ours) we compare our results with various state-of-the-art models.”
ResultImage Restoration: From Sparse and Low-Rank Priors to Deep Priors [Lecture Notes]
Lei Zhang, A., & Zuo, W. (2017) · IEEE Signal Processing Magazine · doi:10.1109/msp.2017.2717489 · Cited by 114
Self-citationBlind Image Deblurring via Deep Discriminative Priors
Li, L., Pan, J., Lai, W., Gao, C., Sang, N., & Yang, M. (2019) · International Journal of Computer Vision · doi:10.1007/s11263-018-01146-0 · Cited by 115
“We note several approaches have been developed to train deepCNNs as image priors or denoisers for non-blind deconvolution (Sreehari et al. 2016; Zhang et al. 2017a, b; Bigdeli et al. 2017).”
Learning Deep Gradient Descent Optimization for Image Deconvolution
Gong, D., Zhang, Z., Shi, Q., van den Hengel, A., Shen, C., & Zhang, Y. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.2968289 · Cited by 134
A Fast Defogging Image Recognition Algorithm Based on Bilateral Hybrid Filtering
Liang, W., Long, J., Li, K., et al. (2020) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3391297 · Cited by 118
“The reported research can be divided into two categories, including image enhancement and image restoration, based on different defogging principles [5, 25, 32].”
BackgroundContent-Aware Scalable Deep Compressed Sensing
Chen, B., & Zhang, J. (2022) · IEEE Transactions on Image Processing · doi:10.1109/tip.2022.3195319 · Cited by 126
Image Restoration Using Joint Patch-Group-Based Sparse Representation
Zha, Z., Yuan, X., Wen, B., Zhang, J., Zhou, J., & Zhu, C. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.3005515 · Cited by 111
“We first report the performance of the proposed JPG-SR for image inpainting and compare it with several leading methods, including SALSA [40], BPFA [28], IPPO [41], ISD-SB [42], JSM [43], Aloha [44], NGS [45], BKSVD [46], TSLRA [47], IRCNN [48] and IDBP [49] methods.”
BackgroundMethodInfluentialRegularization by Denoising via Fixed-Point Projection (RED-PRO)
Cohen, R., Elad, M., & Milanfar, P. (2021) · SIAM Journal on Imaging Sciences · doi:10.1137/20m1337168 · Cited by 118
“ns [70,81,63,43,88,1,82], and it has been extended to other proximal algorithms such as proximal gradient method (PGM) [8,59], approximate message passing (AMP) [58,33,6] and half quadratic splitting [86]. Schemes similar to PnP has been proposed in [28] and [34], where the former is based on the augmented Lagrangian…”
BackgroundMethodMemory-Augmented Deep Unfolding Network for Compressive Sensing
Song, J., Chen, B., & Zhang, J. (2021) · ACM International Conference on Multimedia (ACM MM) · doi:10.1145/3474085.3475562 · Cited by 131
“The models of ReconNet [16], ISTA-Net+ [40], DPA-Net [32], IRCNN [46], MAC-Net [2] and iPiano-Net [30] are trained by the same methods and training datasets with the corresponding works, and DPDNN [5] and GDN [8] utilize the same training dataset with our method due to no CS reconstruction task in their original works.”
MethodGodard, C., Matzen, K., & Uyttendaele, M. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01267-0_33 · Cited by 127
“These methods often make use of neural networks that are purely feed-forward [43, 4, 48, 24, 14, 1, 49], recurrent [44], or a hybrid of the two [6].”
BackgroundDeep 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 113
“[45] adopt CNN denoisers to solve image restoration tasks by manually selecting the hyper-parameters in a model-based optimization framework.”
MethodCoIL: Coordinate-Based Internal Learning for Tomographic Imaging
Sun, Y., Liu, J., Xie, M., Wohlberg, B. E., & Kamilov, U. S. (2021) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2021.3125564 · Cited by 152
Iterative 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 110
“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.”
BackgroundMethodInfluentialDeep learning for high-dimensional reliability analysis
Li, M., & Wang, Z. (2019) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2019.106399 · Cited by 117
Huang, 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 116
Learning Maximally Monotone Operators for Image Recovery
Pesquet, J.-C., Repetti, A., Terris, M., & Wiaux, Y. (2021) · SIAM Journal on Imaging Sciences · doi:10.1137/20m1387961 · Cited by 137
“In practice, either σ` ≡ σ > 0 is chosen to be constant during training [67], or σ` is chosen to be a realization of a random variable with uniform distribution in [0, σ], for σ ∈ ]0,+∞[ (w`)16`6L are assumed to be realizations of standard normal i.”
BackgroundMethodGroup Sparsity Residual Constraint With Non-Local Priors for Image Restoration
Zha, Z., Yuan, X., Wen, B., Zhou, J., & Zhu, C. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.3021291 · Cited by 109
“…[74], Aloha [19], GSR [7], ISD-SB [75], NGS [76], BKSVD [77], JPG-SR [83], TSLRA [84], IR-CNN [24] and IDBP [85] [72], BPFA [16], IPPO [73], JSM [74], A LOHA [19], GSR [7], ISD-SB [75], NGS [76], BKSVD [77], JPG-SR [83], TSLRA [84], IR-CNN [24], IDBP [85] AND GSRC-NLP FOR I MAGE I NPAINTING Fig.”
BackgroundPixelRL: Fully Convolutional Network With Reinforcement Learning for Image Processing
Furuta, R., Inoue, N., & Yamasaki, T. (2019) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2019.2960636 · Cited by 111
“Similar to [29], we added 4,774 images from Waterloo exploration database [56] to the training set.”
BackgroundMethodResultInfluentialDeep 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 111
“The PnP framework has shown great empirical success on diverse applications including denoising [1,52] , restoration [53] , tomography, super-resolution, and fusion.”
BackgroundA Comprehensive Survey of Transformers for Computer Vision
Jamil, S., Piran, M. J., & Kwon, O. (2023) · Drones · doi:10.3390/drones7050287 · Cited by 128
In-Depth Review of YOLOv1 to YOLOv10 Variants for Enhanced Photovoltaic Defect Detection
Hussain, M., & Khanam, R. (2024) · Solar · doi:10.3390/solar4030016 · Cited by 131
“This field encompasses diverse subareas, such as object detection [42], image restoration [43], scene recognition [44], pose and motion estimation [45], object segmentation [46], and video tracking [47].”
PMBANet: Progressive Multi-Branch Aggregation Network for Scene Depth Super-Resolution
Ye, X., Sun, B., Wang, Z., Yang, J., Xu, R., Li, H., & Li, B. (2020) · IEEE Transactions on Image Processing · doi:10.1109/tip.2020.3002664 · Cited by 108
“Meanwhile, R f ( x ) is computed through least squares or using iterative back-projection to refine the result x t without explicitly computing the inversion of K [12], [64].”
BackgroundMethodInfluentialComprehensive and Delicate: An Efficient Transformer for Image Restoration
Zhao, H., Gou, Y., Li, B., Peng, D., Lv, J., & Peng, X. (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/cvpr52729.2023.01357 · Cited by 114
“Method DnCNN [62] IRCNN [63] FFDNet [64] NLRN [24] FOCNet [14] RNAN [66] MWCNN [26] DRUNet [60] SwinIR [22] Ours Ours* #Params 0.”
BackgroundMethodInfluentialDeep Kalman Filtering Network for Video Compression Artifact Reduction
Lu, G. S., Ouyang, W., Xu, D., Zhang, X., Gao, Z., & Sun, M. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01264-9_35 · Cited by 101
Improving Image Restoration by Revisiting Global Information Aggregation
Chu, X., Chen, L., Chen, C., & Lu, X. (2022) · Lecture notes in computer science · doi:10.1007/978-3-031-20071-7_4 · Cited by 117
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 104
“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…”
BackgroundSelf-citationImage denoising in the deep learning era
Izadi, S., Sutton, D. B., & Hamarneh, G. (2022) · Artificial Intelligence Review · doi:10.1007/s10462-022-10305-2 · Cited by 120
“The clean estimate is computed as: DnCNN has been successfully employed in many model-based denoising algorithms serving as an implicit natural image prior (Meinhardt et al. 2017; Zhang et al. 2017).”
BackgroundMethodInfluentialSpectral super-resolution meets deep learning: Achievements and challenges
He, J., Yuan, Q., Li, J., Xiao, Y., Liu, D., Shen, H., & Zhang, L. (2023) · Information Fusion · doi:10.1016/j.inffus.2023.101812 · Cited by 116
“Many works that use the physical model to build a network structure have been conducted [108,109].”
Self-citationReal-World Image Denoising with Deep Boosting
Chen, C., Xiong, Z., Tian, X., Zha, Z.-J., & Wu, F. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2921548 · Cited by 104
Image Restoration via Reconciliation of Group Sparsity and Low-Rank Models
Zha, Z., Wen, B., Yuan, X., Zhou, J., & Zhu, C. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3078329 · Cited by 100
“We also compare the proposed LR-GSC with deep learning-based image deblocking approaches, including AR-CNN [60], IRCNN [61] and FFDNet [62].”
MethodImage 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 109
“expand their previous work [42] and propose a plug-and-play unrolled model using deep denoiser prior for image restoration.”
BackgroundMethodInfluentialNTIRE 2019 Challenge on Real Image Denoising: Methods and Results
Abdelhamed, A., Timofte, R., Brown, M. S., et al. (2019) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw.2019.00273 · Cited by 85
“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].”
BackgroundSelf-citationDynamic Attentive Graph Learning for Image Restoration
Mou, C., Zhang, J., & Wu, Z. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00429 · Cited by 105
“In this task, we compare our method with RNAN [52] and IRCNN [50] on McMaster18 [50], Kodak24, and Urban100 test sets.”
BackgroundMethodInfluentialSRConvNet: 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 113
“WeadopttheDIV2K(Tim-ofte et al., 2017) dataset for training and test our method on CBSD64 (Arbeláez et al., 2011), Kodak24 (Zhang et al., 2017), and McMaster (Zhang et al., 2011).”
InfluentialSimultaneous Fidelity and Regularization Learning for Image Restoration
Ren, D., Zuo, W., Zhang, D. D., Lei Zhang, A., & Yang, M. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2926357 · Cited by 98
Self-citationNonlocal CNN SAR Image Despeckling
Cozzolino, D., Verdoliva, L., Scarpa, G., et al. (2020) · Remote Sensing · doi:10.3390/rs12061006 · Cited by 101
“Then, in [25], homomorphic filtering is performed based on multiple instances of the same CNN [47] trained on Gaussian noise at various levels of intensity.”
MethodDeep 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 104
CoRRN: Cooperative Reflection Removal Network
Wan, R., Shi, B., Li, H., Duan, L., Tan, A., & Kot, A. C. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2921574 · Cited by 101
“Previous works that use deep learning to solve the inverse imaging problems [45], [46] or layer separation problems [20] mainly optimize the whole network on patches with resolution n n cropped from the whole images.”
BackgroundDeep Blind Video Super-resolution
Pan, J., Bai, H., Dong, J., Zhang, J., & Tang, J. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00477 · Cited by 102
“Motivated by the success of [40, 41], we first estimate an intermediate HR image by a simple image deconvolution model and then explore the information of adjacent frames and deep CNN models to restore high-quality images.”
BackgroundDing, F., Yu, K., Gu, Z., et al. (2021) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2021.3120075 · Cited by 98
InstructIR: High-Quality Image Restoration Following Human Instructions
Conde, M. V., Geigle, G., & Timofte, R. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72764-1_1 · Cited by 144
“Some methods focus on specific degradations, for instance reducing noise (denoising) [64, 102, 103], removing blur (deblurring) [58, 105], or clearing haze (dehazing) [16, 66].”
BackgroundMethodScalable Plug-and-Play ADMM With Convergence Guarantees
Sun, Y., Wu, Z., Xu, X., Wohlberg, B. E., & Kamilov, U. S. (2021) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2021.3094062 · Cited by 98
COLA-Net: Collaborative Attention Network for Image Restoration
Mou, C., Zhang, J., Xiaopeng Fan, B., Liu, H., & Wang, R. (2021) · IEEE Transactions on Multimedia · doi:10.1109/tmm.2021.3063916 · Cited by 111
“To aggregate more useful information from a large receptive field, some strategies such as hourglass-shaped architecture [8], [9], [10], [11], dilated convolutions [12], [13], [6] or stacking more convolutional layers [14], [7], [15], [16] are applied in image restoration tasks.”
MethodInfluentialBlind image super-resolution with spatially variant degradations
Cornillère, V., Djelouah, A., Wang, Y., Sorkine‐Hornung, O., & Schroers, C. (2019) · ACM Transactions on Graphics · doi:10.1145/3355089.3356575 · Cited by 90
“This is the case of [Rick Chang et al. 2017; Zhang et al. 2017] that show how a deep CNN trained for image denoising can effectively be used as prior in various image restoration tasks including SR.”
MethodDilated Deep Residual Network for Image Denoising
Wang, T., Sun, M., & Hu, K. (2017) · IEEE International Conference on Tools with Artificial Intelligence (ICTAI) · doi:10.1109/ictai.2017.00192 · Cited by 111
“Besides Jain’s work [5] and MLP [6], successful methods in this category also include CSF[7], DGCRF [8], NLNet [9], TNRD [10], and DP [11].”
BackgroundMethodInfluentialTensor Completion via Complementary Global, Local, and Nonlocal Priors
Zhao, X.-L., Yang, J.-H., Ma, T.-H., Jiang, T.-X., Ng, M. K., & Huang, T. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3138325 · Cited by 96
“For various image recovery tasks, Zhang et al. [69] learned the deep denoising prior expressed by a CNN.”
MethodDWDN: Deep Wiener Deconvolution Network for Non-Blind Image Deblurring
Dong, J., Roth, S., & Schiele, B. (2021) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2021.3138787 · Cited by 96
“The images deblurred by the methods [3], [6], [15], [17], [18], [45], [50], [57] contain severe artifacts as shown in (b) and (d)–(j) .”
BackgroundMethodInfluentialMop moiré patterns using mopnet
He, B., Wang, C., Shi, B., & Duan, L. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00251 · Cited by 97
“Also, general degradation oriented universal learning framework has been investigated, with prior of degradation kernel [28] or exploitation of self-similarity through non-local operation [14].”
MethodMemory-augmented Deep Conditional Unfolding Network for Pansharpening
Yang, G., Zhou, M., Yan, K., Liu, A., Fu, X., & Wang, F. (2022) · 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/cvpr52688.2022.00183 · Cited by 92
3D Common Corruptions and Data Augmentation
Kar, O. F., Yeo, T., Atanov, A., & Zamir, A. (2022) · 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/cvpr52688.2022.01839 · Cited by 106
“Image restoration aims to undo the corruption in the image using classical signal processing techniques [18, 20, 34,41] or learning-based approaches [1,8,45,46,56,83,84].”
BackgroundA Deep Journey into Super-resolution: A surveyPreprint
Anwar, S., Khan, S. H., & Barnes, N. M. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.07523 · Cited by 86
“The architecture of SRMD is similar to [28], [30], [31].”
BackgroundResultInfluentialLearning to Optimize: A Primer and A BenchmarkPreprint
Chen, T., Chen, X., Chen, W., Heaton, H., Liu, J., Wang, Z., & Yin, W. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2103.12828 · Cited by 129
“Th works [175; 153; 76; 256]use data-driven training to learn a proximal or a projection operator and apply it in the unrolled algorithm.”
BackgroundMethodResultInfluentialMulti-scale Weighted Nuclear Norm Image Restoration
Yair, N., & Michaeli, T. (2018) · 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.2018.00334 · Cited by 94
“2): IRCNN [49] which uses a CNN denoiser, and RED [38] with TNRD [11] as the denoising engine.”
BackgroundMethodInfluentialMultilevel Edge Features Guided Network for Image Denoising
Fang, F., Li, J., Yuan, Y., Zeng, T., & Zhang, G. (2020) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2020.3016321 · Cited by 87
“In order to verify the effectiveness of MLEFGN, we compare MLEFGN with more than 14 image denoising methods, including EPLL [54], BM3D [15], MLP [20], RED [56], MemNet [57], TNRD [43], NLED 6 7 × 7 [44], NN3D [58], WNNM [16], IRCNN [45], DnCNN [26], FFDNet [28], ADNet [46], and N 3 Net [55].”
MethodContext Reasoning Attention Network for Image Super-Resolution
Zhang, Y., Wei, D., Qin, C., Wang, H., Pfister, H., & Fu, Y. (2021) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv48922.2021.00424 · Cited by 98
“We further provide comparisons under the more challenging DN degradation model [61, 66], where the LR images are further added with heavy noise (noise level=30).”
BackgroundMethodInfluentialUnpaired Deep Learning for Accelerated MRI Using Optimal Transport Driven CycleGAN
Oh, G., Sim, B., Chung, H., Sunwoo, L., & Ye, J. C. (2020) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2020.3018562 · Cited by 81
“In contrast to the recent inverse problem approaches using deep learning prior [42], [43], which incorporate physics-driven information during the run-time reconstruction, our cycleGAN formulation shows that the physics-driven constraint can be incorporated in the training phase of a feed-forward deep neural networks…”
ResultImage De-Noising With Machine Learning: A Review
Thakur, R. S., Chatterjee, S., Yadav, R. N., & Gupta, L. (2021) · IEEE Access · doi:10.1109/access.2021.3092425 · Cited by 92
EWT: Efficient Wavelet-Transformer for single image denoising
Li, J., Cheng, B., Chen, Y., Gao, G., Shi, J., & Zeng, T. (2024) · Neural Networks · doi:10.1016/j.neunet.2024.106378 · Cited by 92
Metadata
- Authors
- Zhang, K., Zuo, W., Gu, S., & Zhang, L.
- Affiliations
- Hong Kong Polytechnic University (HK)
- Harbin Institute of Technology (CN)
- Venue
- 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2017-07
- Pages
- 2808-2817
- OpenAlex
- W2613155248
- Cited by
- 2,116
- References
- 81
- Citations by year
- 2026: 80 · 2025: 191 · 2024: 248 · 2023: 293 · 2022: 272 · 2021: 338 · 2020: 297 · 2019: 240 · 2018: 135 · 2017: 10
- FWCI
- 49.67
- Type
- conference paper
- Language
- en
- Open access
- green
- hdl.handle.net
- Locations
- IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
- PolyU Institutional Research Archive (Hong Kong Polytechnic University) (submittedVersion)
- TUScholarShare (Temple University) (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Image and Signal Denoising Methods
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Image Processing Techniques
- Physical Sciences › Engineering › Media Technology › Advanced Image Fusion Techniques
- Keywords
- image restoration, model-based optimization, discriminative learning, variable splitting, inverse problems, low-level vision, Gaussian denoising, image deblurring
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.
A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
