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
25 Own interpretation · 16 Own result · 10 No basis given · 9 No full text of the sources · 3 Sources not found · 3 Own method · 2 Not checked yet · 1 Supported · 1 Not supported · 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- 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
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- 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.
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a 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
“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.
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
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
“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
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a 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
“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.
plainconditional- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
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 the proposed denoiser is very competitive in both CPU and GPU implementation.”
§ Learning Deep CNN Denoiser Prior for Image Restoration > 4.1. Image Denoising
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a 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
“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.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says 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
“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
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
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 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- 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)
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plainn = 2- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plainSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plaindirectSingle case- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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?
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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?
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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?
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably 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.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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
plaindirectn = 2- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably 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
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably 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
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It 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.”
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 1,001–1,100 of 2,094
CEPDNet: a fast CNN-based image denoising network using edge computing platform
Bai, X., Wan, Y., & Wang, W. (2024) · The Journal of Supercomputing · doi:10.1007/s11227-024-06646-0 · Cited by 8
“…in preserving edge details, we compared the proposed method CEPDNet with BM3D [4], WNNM [5], DnCNN [7], FFDNet [8], ADNet [12], EPLL [22], IRCNN [23], TNRD [24], DudeNet [25], MDRN [26], TWSC [27], RIDNet [28], CBDNet [9], AINDNet [29], SADNet [30], BRDNet [13], DualBDNet [31], CDNet [32], RDN…”
NG-RED:Nonconvex group-matrix residual denoising learning for image restoration
Li, Y., Wu, H., Jiang, X., & Ding, X. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.125876 · Cited by 8
$\mathrm{D}^{3}$T: Deep Denoising Dictionary Tensor for Hyperspectral Anomaly Detection
Shen, Q., Liu, Z., Wang, H., Xu, Y., Chen, Y., & Liang, Y. (2024) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2024.3523133 · Cited by 8
“For example, Zhang et al. [54], [55], [56] designed a series of deep denoisers for image restoration.”
Deep Joint Unrolling for Deblurring and Low-Light Image Enhancement (JUDE)
Vo, T., & Park, C. Y. (2025) · IEEE Winter Conference on Applications of Computer Vision · doi:10.1109/wacv61041.2025.00267 · Cited by 8
“Traditional methods include Lee et al.’s approach [17], which employs a tree-like layered structure to represent gray-level differences and solves a constrained optimization problem to determine the transformation function.”
WaveUIR: wavelet-based guided transformer model for efficient universal image restoration
Zhu, Z., Xiong, B., Rao, Z., Zhu, J., Li, W., Ou, Q., & Li, X. (2025) · The Visual Computer · doi:10.1007/s00371-025-03895-9 · Cited by 10
AGEM: Solving Linear Inverse Problems via Deep Priors and Sampling
Guo, B., Han, Y., & Wen, J. (2019) · neural information processing systems · Cited by 9
DenoisingNet: An Efficient Convolutional Neural Network for Image Denoising
Li, Y., Miao, Z., Zhang, R., & Wang, J. (2019) · International Conference on Artificial Intelligence and Big Data (ICAIBD) · doi:10.1109/icaibd.2019.8837023 · Cited by 8
Image restoration model using Jaya‐Bat optimization‐enabled noise prediction map
Premnath, S. P., & Renjit, J. A. (2021) · IET Image Processing · doi:10.1049/ipr2.12162 · Cited by 8
“The DCNN classifier [27] generates the prediction map based on input image.”
BackgroundMethodA Dual Model for Restoring Images Corrupted by Mixture of Additive and Multiplicative Noise
Zhao, C., Liu, J., & Zhang, J. (2021) · IEEE Access · doi:10.1109/access.2021.3137995 · Cited by 8
CNN-based denoising system for the image quality enhancement
Kumar, S. P. P., & Kurmi, Y. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-12406-0 · Cited by 8
“Figures 4, 5 and 6 represent the visual effect of proposed method along with qualitative performance of the different methods as BM3B [9], WNNM [18], TNRD [7], IDCNN [42], DnCNN [43], IRCNN [44], datasets SET12 and BSD68 at a higher noise density level ( ‘ σ = 50 ’ ).”
BackgroundMethodInfluentialA Conspectus of Deep Learning Techniques for Single-Image Super-Resolution
Pandey, G., & Ghanekar, U. (2022) · Pattern Recognition and Image Analysis · doi:10.1134/s1054661822010059 · Cited by 8
“Over the years, different NE-based techniques are proposed such as non-negative NE [12], partially supervised NE [194], NE with self-similar image prior [205], low rank matrix recovery NE [21], sparse NE [37], anchored NE [160], partially supervised anchored NE [150], and rotation expanded…”
BackgroundBAM: a balanced attention mechanism to optimize single image super-resolution
Wang, F., Hu, H., Shen, C., et al. (2022) · Journal of Real-Time Image Processing · doi:10.1007/s11554-022-01235-x · Cited by 8
Adaptive feature denoising based deep convolutional network for single image super-resolution
Cheng, R., Wu, Y., Wang, J., Ma, M., Niu, Y., & Shi, G. (2022) · Computer Vision and Image Understanding · doi:10.1016/j.cviu.2022.103518 · Cited by 8
Qiao, X., Huang, Y., & Li, W. (2023) · Magnetic Resonance in Medicine · doi:10.1002/mrm.29575 · Cited by 8
“29,50,51 Compared with them, our regularizer directly learn from enriched information to obtain better results.”
BackgroundMultisource Information Fusion Network for Optical Remote Sensing Image Super-Resolution
Shi, M., Gao, Y., Chen, L., et al. (2023) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2023.3242039 · Cited by 8
“Then, the methods of deep unfolding for image SR, deblurring, and denoising have been proposed [44], [45], [46].”
MethodYoung, S. I., Dalca, A. V., Ferrante, E., Golland, P., Metzler, C. A., Fischl, B. R., & Iglesias, J. E. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3299789 · Cited by 9
“Reehorst and Schniter [28] clarify the conditions on such denoisers to guarantee convergence while [46] demonstrate the utility of learned denoisers for regularization.”
CMID: Crossmodal Image Denoising via Pixel-Wise Deep Reinforcement Learning
Guo, Y., Gao, Y., Hu, B., Qian, X., & Liang, D. (2023) · Sensors · doi:10.3390/s24010042 · Cited by 8
“Due to the variability of the external environment and inherent limitations of a device, noise inevitably occurs during image-sensing processes [1,2], which can seriously reduce the visual quality of the image and adversely affect downstream computer vision and multimedia tasks [3].”
BackgroundScene recovery: Combining visual enhancement and resolution improvement
Zhang, H., Qi, T., & Zeng, T. (2024) · Pattern Recognition · doi:10.1016/j.patcog.2024.110529 · Cited by 10
A Multiscale Dual Attention Sparse Dual Network for Time-Frequency Image Denoising of Radar Signal
Li, S., Du, X., Cui, G., Chen, X., & Zheng, J. (2024) · IEEE Sensors Journal · doi:10.1109/jsen.2024.3403856 · Cited by 8
“…that MDASDNet produces satisfactory outcomes at the low SNR. 2) Comparison With Representative Denoising Methods: In this section, some representative denoising methods are investigated to verify the advantage of MDASDNet, including DnCNN [18], IRCNN [20], ADNet [22], DudeNet [24], and BRDNet [23].”
BackgroundMethodCarneiro-Esteves, S., Vacavant, A., & Merveille, O. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.128055 · Cited by 8
“Following this work, a series of works [16, 54, 55, 56] has been proposed to replace the regularization term by a denoising neural network, which learns the noise distribution of a dataset of interest.”
Attention-Enhanced Convolutional Iterative Unrolling Network for Under-Sampled MRI Reconstruction
Ma, M., & Gan, H. (2024) · IEEE Transactions on Emerging Topics in Computational Intelligence · doi:10.1109/tetci.2024.3437171 · Cited by 8
“In recent years, Deep Learning (DL) has achieved excellent results in image recovery problems, such as denoising [19], [20], deblurring [21], [22], and super-resolution [23], [24].”
Restoration of turbulence-degraded images based on deep convolutional network
Bai, X., Liu, M., He, C., et al. (2019) · SPIE Applications of Machine Learning · doi:10.1117/12.2527593 · Cited by 8
“4 Network Architecture Inspired by SRCNN[24] and IRCNN[25], we investigate a FCN architectures with residual blocks and dilated convolutions.”
MethodInfluentialYang, H., Su, X., & Chen, S. (2020) · Applied Sciences · doi:10.3390/app10072437 · Cited by 8
“trained a set of convolutional neural network priors, which act as a constraint of model-based optimization methods to cope with the image restoration problems [25].”
BackgroundA Hybrid Structural Sparse Error Model for Image Deblocking
Zha, Z., Yuan, X., Zhou, J., et al. (2020) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp40776.2020.9053968 · Cited by 8
“Note that the proposed HSSE is a general model, and therefore, we can use the other methods to replace SSR to solve β sub-problem, such as deep CNN denoisier [29].”
MethodImage Restoration via Deep Memory-Based Latent Attention Network
Zhang, X., Gao, P., Zhao, K., et al. (2020) · IEEE Access · doi:10.1109/access.2020.2999965 · Cited by 8
Residual Forward-Subtracted U-Shaped Network for Dynamic and Static Image Restoration
Jung, H. M., Kim, B. H., & Kim, M. Y. (2020) · IEEE Access · doi:10.1109/access.2020.3011580 · Cited by 8
An Improved Combination of Image Denoisers Using Spatial Local Fusion Strategy
Liu, Y., Xu, S., & Lin, Z. (2020) · IEEE Access · doi:10.1109/access.2020.3016766 · Cited by 8
A Novel Learned Primal-Dual Network for Image Compressive Sensing
Zhang, C., Liu, Y., Shang, F., et al. (2021) · IEEE Access · doi:10.1109/access.2021.3057621 · Cited by 8
Image Denoising Using Nonlocal Regularized Deep Image Prior
Xie, Z., Liu, L., Luo, Z., & Huang, J. (2021) · Symmetry · doi:10.3390/sym13112114 · Cited by 8
“, FFDNet [41] and IRCNN [26], which have been shown to be superior to the well-known benchmark method, i.”
MethodImage denoising based on global image similar patches searching and HOSVD to patches tensor
Guo, J., Chen, H., Shen, Z., et al. (2021) · EURASIP Journal on Advances in Signal Processing · doi:10.1186/s13634-021-00798-4 · Cited by 8
“There are many different arrangement ways of the nth-mode matricization, but in practice, permutation of elements doesn’t affect the result of calculation [27].”
BackgroundXu, W., Zhou, Y., Liu, D., Wang, X., & Chen, W. (2022) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2022.3166029 · Cited by 8
Hybrid Deep Learning Framework for Reduction of Mixed Noise via Low Rank Noise Estimation
Kim, D.-G., Ali, Y., Farooq, M. A., et al. (2022) · IEEE Access · doi:10.1109/access.2022.3170490 · Cited by 8
Lq -SPB-Net: A Real-Time Deep Network for SAR Imaging and Despeckling
Xiong, K., Zhao, G., Wang, Y., et al. (2021) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2021.3087610 · Cited by 8
Zhuang, X. D., Zhou, D., Cai, J., et al. (2023) · IEEE Sensors Journal · doi:10.1109/jsen.2023.3293136 · Cited by 8
“Besides that, in order to verify the effectiveness and superiority of HYDP, image restoration CNN (IRCNN) [44], and derivative alternating direction method of multipliers (D-ADMM) [45] are selected as benchmarks for comparison.”
BackgroundInfluentialOn the Contractivity of Plug-and-Play Operators
Athalye, C. D., Chaudhury, K. N., & Kumar, B. (2023) · IEEE Signal Processing Letters · doi:10.1109/lsp.2023.3323248 · Cited by 8
“We notice that the reconstructions are of high quality, as already reported in the literature [6], [10], [17].”
BackgroundResultLiu, R., Liu, Z., Mu, P., Fan, X., & Luo, Z. (2023) · IEEE Transactions on Image Processing · doi:10.1109/tip.2023.3328486 · Cited by 8
“Various CNN modules [9], [10], [11] have been directly integrated into numerical optimization iterations as the implicitly-defined data-driven priors, such as CNN-based denoisers for deconvolution [11], Retinex-based modules for low-light enhancement [10], and discriminative classifiers for…”
MethodInfluentialMultiplicative noise removal using deep CNN denoiser prior
Wang, G., Wang, G., Pan, Z., et al. (2017) · International Symposium on Intelligent Signal Processing and Communication Systems (ISPACS) · doi:10.1109/ispacs.2017.8265635 · Cited by 8
Learning an Efficient Convolution Neural Network for Pansharpening
Guo, Y., Ye, F., & Gong, H. (2019) · Algorithms · doi:10.3390/a12010016 · Cited by 8
“As a trade-off between receptive field and network complexity, we adopt dilated convolution [19,20] as a substitute for traditional convolutional layer.”
MethodSINGLE CHANNEL JOINT SPEECH DEREVERBERATION AND DENOISING USING DEEP PRIORS
Raikar, A., Basu, S., & Hegde, R. M. (2018) · IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2018.8646327 · Cited by 8
“It has been shown in [8] that this term can instead be alternatively solved by using a denoiser.”
BackgroundMethodBlind Image Deblurring Based on Local Edges Selection
Han, Y., & Kan, J. (2019) · Applied Sciences · doi:10.3390/app9163274 · Cited by 8
“Some methods cannot obtain clear images successfully, the image is still suffered from serious blur, such as [1,41]; In some deblurring images, such as the images estimated by [7,19,20,22], ringing artifacts affect the deblurred results.”
BackgroundMethodInfluentialEdge‐aware image filtering using a structure‐guided CNN
Kim, S.-J., Song, C., Jang, J., et al. (2019) · IET Image Processing · doi:10.1049/iet-ipr.2018.6691 · Cited by 8
Image Super-Resolution Based on CNN Using Multilabel Gene Expression Programming
Tang, J., Huang, C., Liu, J., & Zhu, H. (2020) · Applied Sciences · doi:10.3390/app10030854 · Cited by 8
Robust Prior-Based Single Image Super Resolution Under Multiple Gaussian Degradations
Wang, W., Wu, G., Cai, W., Zeng, L., & Chen, J. (2020) · IEEE Access · doi:10.1109/access.2020.2987911 · Cited by 8
Learning a Multi-scale Deep Residual Network of Dilated-Convolution for Image Denoising
Chen, G., Gao, Z., Zhu, P., & Chen, Z. J. (2020) · International Conference on Cloud Computing and Big Data Analytics (ICCCBDA) · doi:10.1109/icccbda49378.2020.9095754 · Cited by 11
“This article's reference methods can be classified into model-based algorithms such as BM3D [5], deep learning algorithms such as DnCNN [9], CNDNet [12], S_dilated [27], and non-local algorithms such as N 3 Net [15], GraphCNN [16].”
MethodDeep Gradient Prior Regularized Robust Video Super-Resolution
Song, Q., & Liu, H. (2021) · Electronics · doi:10.3390/electronics10141641 · Cited by 8
From Local to Global: Efficient Dual Attention Mechanism for Single Image Super-Resolution
Zhang, P., & Lam, E. Y. M. (2021) · IEEE Access · doi:10.1109/access.2021.3105726 · Cited by 8
Simultaneous Fidelity and Regularization Learning for Image RestorationPreprint
Ren, D., Zuo, W., Zhang, D. D., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.04522 · Cited by 8
Self-citationSelf-Supervised Fast Adaptation for Denoising via Meta-LearningPreprint
Lee, S., Cho, D., Kim, J., & Kim, T. H. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2001.02899 · Cited by 8
“[32] adopted deep neural networks for denoising and inpainting tasks, numerous follow-up studies have been proposed [34, 35, 36, 38, 20, 23, 37, 17, 2].”
BackgroundMethodLearning Deep Gradient Descent Optimization for Image DeconvolutionPreprint
Gong, D., Zhang, Z., Shi, Q., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1804.03368 · Cited by 8
Non-Local Part-Aware Point Cloud DenoisingPreprint
Huang, C., Li, R., Li, X., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2003.06631 · Cited by 8
“…part-aware denoising; and (iii) noise component N generally has small magnitude and simple patterns, so we extract the latent noise map progressively from the noisy inputs, instead of directly estimating the noise-free underlying geometric structures, similar to routines in image denoising [35,36].”
A Comprehensive Review of Deep Learning-based Single Image Super-resolutionPreprint
Bashir, S. M. A., Wang, Y., Khan, M., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.09351 · Cited by 8
“In dilated convolution, the contextual information is used to generate realistic-looking SR images as in [129], dilated convolution was used to double the receptive field, which resulted in better results.”
MethodDevice-edge collaborative occluded face recognition method based on cross-domain feature fusion
Zhang, P., Tan, L., Yang, Z., Huang, F., Sun, L., & Peng, H. (2024) · Digital Communications and Networks · doi:10.1016/j.dcan.2024.05.003 · Cited by 8
Frequency-Aware Divide-and-Conquer for Efficient Real Noise Removal
Huang, Y., Liu, C., Li, B., et al. (2024) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2024.3439591 · Cited by 7
High-Throughput Decomposition-Inspired Deep Unfolding Network for Image Compressed Sensing
Li, T., Yan, Q., Li, Y., et al. (2025) · IEEE Transactions on Computational Imaging · doi:10.1109/tci.2025.3527880 · Cited by 7
“DUNs arebuiltupontheoptimizationalgorithms,includingbutnot limitedtotheProximalGradientDescent(PGD)algorithm, ApproximateMessagePassing(AMP)algorithm,Alternating DirectionMethodofMultipliers(ADMM)algorithm,andHalf QuadraticSplitting(HQS)[18]algorithm.”
FRANet: A Feature Refinement Attention Network for SAR Image Denoising
Liu, S., Lei, Y., Hu, Q., et al. (2025) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2025.3564846 · Cited by 7
Underwater image restoration via transmission and Wasserstein distance constraints
Gao, B., Feng, X., Wang, K., & Zhu, H. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103580 · Cited by 7
SAR Imaging and Despeckling Based on Sparse, Low-Rank, and Deep CNN Priors
Xiong, K., Zhao, G., Wang, Y., et al. (2021) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2021.3131201 · Cited by 7
Imaging Domain Seismic Denoising Based on Conditional Generative Adversarial Networks (CGANs)
Zhang, H., & Wang, W. (2022) · Energies · doi:10.3390/en15186569 · Cited by 7
Zhang, D., Lian, Q., Su, Y., & Ren, T. (2023) · IEEE/CAA Journal of Automatica Sinica · doi:10.1109/jas.2023.123390 · Cited by 7
“Therefore, a stepwise approximation of the Bisection method [40] is adopted to seek a numerical solution.”
MethodConstrained Plug-and-Play Priors for Image Restoration
Benfenati, A., & Cascarano, P. (2024) · Journal of Imaging · doi:10.3390/jimaging10020050 · Cited by 7
“As discussed in the previous section, we adopt the DnCNN introduced in [47] as the denoising prior due to its state-of-the-art performance in the field and its fast computation.”
MethodInfluentialLow-Light Image Restoration Using a Convolutional Neural Network
Hussain, S. A., Chalicham, N., Garine, L., et al. (2024) · Journal of Electronic Materials · doi:10.1007/s11664-024-11079-9 · Cited by 7
“10,17 Furthermore, CNN designs may be customized to incorporate contextual information and hierarchical characteristics, speeding the restoration process by making efficient use of both local and global image context.”
STVDNet: spatio-temporal interactive video de-raining network
Ouyang, Z., Zhao, H., Zhang, Y., et al. (2024) · The Visual Computer · doi:10.1007/s00371-024-03565-2 · Cited by 7
Unsupervised super resolution using dual contrastive learning
Wu, C. Q., & Jing, Y. (2025) · Neurocomputing · doi:10.1016/j.neucom.2025.129649 · Cited by 7
Denoising x-vectors for Robust Speaker Recognition
Mohammadamini, M., Matrouf, D., & Noé, .-G. (2020) · The Speaker and Language Recognition Workshop (Odyssey) · doi:10.21437/odyssey.2020-11 · Cited by 7
“A. Gaussian DAE: Adding a regularization term to the loss function is a common way to constraint the solution space in denoising DNNs [16, 17].”
BackgroundRegularization via Deep Generative Models: an Analysis Point of View
Oberlin, T., & Verm, M. (2021) · Proceedings - International Conference on Image Processing · doi:10.1109/icip42928.2021.9506138 · Cited by 7
“Some works proposed to learn a denoising networks [10, 11] and use it as a plug-and-play prior; or to learn a 1-class classification DNN [12] which is used as a projection operator onto the set of natural images.”
BackgroundSun, D., Shi, Y., & Feng, Y. (2020) · IET Image Processing · doi:10.1049/iet-ipr.2020.1193 · Cited by 7
“In [23, 24], neural network denoising achieved good results.”
BackgroundAn Approach to Image Denoising Using Autoencoders and Spatial Filters for Gaussian Noise
Patil, A. P., Pramod, A., Harish, A., et al. (2021) · International Conference on Cloud Computing, Data Science & Engineering (Confluence) · doi:10.1109/confluence51648.2021.9377166 · Cited by 7
Lei, J., & Liu, Q. (2021) · Expert Systems with Applications · doi:10.1016/j.eswa.2021.116296 · Cited by 7
Improved Deep Multi-Patch Hierarchical Network For Handling Saturation In Image Deblurring
Mahendra, B. M., Sonoli, S., & Ameta, A. (2022) · Array · doi:10.1016/j.array.2022.100228 · Cited by 7
Multi-Scale Detail Enhancement Network for Image Super-Resolution
Wang, S., & Zheng, J. Y. (2022) · Proceedings - International Conference on Pattern Recognition/Proceedings/International Conference on Pattern Recognition · doi:10.1109/icpr56361.2022.9956541 · Cited by 7
EFID: Edge-Focused Image Denoising Using a Convolutional Neural Network
Holla, K. S., Park, N., & Lee, B. (2023) · IEEE Access · doi:10.1109/access.2023.3239835 · Cited by 7
“[9], (e) FFDNet [10], (f) IRCNN [11], (g) BRDNet [13], (h) ADNet [12], (i) DRUNet [16], (j) MLEFGN [28], and (k) Proposed EFID.”
BackgroundMethodInfluentialA Highly Interpretable Deep Equilibrium Network for Hyperspectral Image Deconvolution
Γκίλλας, Α., Ampeliotis, D., & Berberidis, K. (2023) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49357.2023.10095286 · Cited by 7
“Additionally, sub-problem (5b) can be written as follows Based on Bayesian estimation theory, relation (8) can be be interpreted as a Gaussian denoiser with noise level ( λ/b ) [11].”
BackgroundHigh performance dilated convolutions on multi-core DSPs
Wang, Y., Wang, Q., Pei, X., et al. (2023) · CCF Transactions on High Performance Computing · doi:10.1007/s42514-023-00166-8 · Cited by 7
MFGAN: towards a generic multi-kernel filter based adversarial generator for image restoration
Chahi, A., Kas, M., Kajo, I., & Ruichek, Y. (2023) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-023-01959-7 · Cited by 8
A new multi-scale CNN with pixel-wise attention for image denoising
Deng, J., & Hu, C. (2023) · Signal Image and Video Processing · doi:10.1007/s11760-023-02944-x · Cited by 7
Spectral Superresolution Using Transformer with Convolutional Spectral Self-Attention
Liao, X., He, L., Mao, J., et al. (2024) · Remote Sensing · doi:10.3390/rs16101688 · Cited by 7
Feature-preserving ultrasound speckle reduction via L 0 minimization
Zhu, L., Wang, W., Li, X., et al. (2018) · Neurocomputing · doi:10.1016/j.neucom.2018.03.009 · Cited by 7
SBM3D: Sparse Regularization Model Induced by BM3D for Weighted Diffraction Imaging
Shi, B., Lian, Q., Chen, S., et al. (2018) · IEEE Access · doi:10.1109/access.2018.2865997 · Cited by 7
Joint denoising and decompression using CNN regularization
González, M., Preciozzi, J., Musé, P., et al. (2018) · HAL (Le Centre pour la Communication Scientifique Directe) · Cited by 7
Exponential linear unit dilated residual network for digital image denoising
Panda, A., Naskar, R., & Pal, S. (2018) · Journal of Electronic Imaging · doi:10.1117/1.jei.27.5.053024 · Cited by 6
“This gain value is better than that obtained by most state-of-the-art CNN denoising models such as DRNID6 or IRCNN.7 Improvements in PSNR over state-of-the-art are obtained in case of both Set12 and set BSD68.”
BackgroundMethodInfluentialMultipath feedforward network for single image super-resolution
Shen, M., Yu, P., Wang, R., et al. (2019) · Multimedia Tools and Applications · doi:10.1007/s11042-019-7334-9 · Cited by 7
“For example, VDSR [16] and IRCNN [44] based on the VGG [27] stack more convolutional layers to build model and use a single model accomplish multi-scale image super-resolution.”
MethodImage Denoising via Multi-Scale Gated Fusion Network
Li, S., Chen, Y., Jiang, R., et al. (2019) · IEEE Access · doi:10.1109/access.2019.2910879 · Cited by 7
“As dilated convolution is widely used in semantic segmentation, it has recently become popular for image denoising [14], [21], [22].”
Compounded Layer-Prior Unrolling: A Unified Transmission-Based Image Enhancement Framework
Liu, R., Hou, M., Liu, J., et al. (2019) · IEEE International Conference on Multimedia and Expo (ICME) · doi:10.1109/icme.2019.00099 · Cited by 7
“It is investigated that, from a Bayesian perspective, the prior of regularization term approximately follows Gaussian distribution with a variance related to corresponding parameter [14].”
BackgroundMethodSuper-Resolution on Degraded Low-Resolution Images Using Convolutional Neural Networks
Albluwi, F., Krylov, V. A., & Dahyot, R. (2019) · European Signal Processing Conference (EUSIPCO) · doi:10.23919/eusipco.2019.8903000 · Cited by 7
“compare our model with VDSR [21], SRMD [23], also modelbased methods such as IRCNN [34].”
BackgroundMethodInfluentialGao, Y., Tan, J., Shi, Y., Lu, S., Gupta, A. K., Li, H., & Liang, Z. (2020) · Journal of medical imaging · doi:10.1117/1.jmi.7.3.032502 · Cited by 7
Edge‐guided single facial depth map super‐resolution using CNN
Zhang, F., Liu, N., Chang, L., Duan, F., & Deng, X. (2020) · IET Image Processing · doi:10.1049/iet-ipr.2019.1623 · Cited by 7
“Later, VDSR [20], IRCNN [25] introduced deeper networks with residual learning, where residual learning is to recover the high-frequency component between LR and HR images.”
BackgroundMulti-Task Learning Framework for Motion Estimation and Dynamic Scene Deblurring
Jung, H., Kim, Y., Jang, H., Ha, N., & Sohn, K. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3113185 · Cited by 7
Investigating Customization Strategies and Convergence Behaviors of Task-Specific ADMM
Liu, R., Mu, P., & Zhang, J. (2021) · IEEE Transactions on Image Processing · doi:10.1109/tip.2021.3113796 · Cited by 7
Image Super-Resolution via Dual-Level Recurrent Residual Networks
Tan, C., Wang, L., & Cheng, S. (2022) · Sensors · doi:10.3390/s22083058 · Cited by 7
“Conclusions and Discussion In this paper, we realize image super-resolution reconstruction by adding an extra level in the super-resolution network based on feedforward structure, called super-resolution via dual-level recurrent residual network (DLRRN), which makes the pieces of HRL “butterfly”from Set5 “img_044”from…”
MethodInfluentialXception Network for Weather Image Recognition Based on Transfer Learning
An, N. H. (2022) · International Conference on Machine Learning and Intelligent Systems Engineering (MLISE) · doi:10.1109/mlise57402.2022.00072 · Cited by 7
A Review on Underwater Image Enhancement Techniques
Amala, V. R., & Raj, S. M. A. (2023) · International Conference on Control, Communication and Computing (ICCC) · doi:10.1109/iccc57789.2023.10165194 · Cited by 7
“scene super-resolution was developed in the article [19].”
BackgroundMethodHow Should we Handle 4D Light Fields with CNNS?
Fujita, S., Takahashi, K., & Fujii, T. (2018) · Proceedings - International Conference on Image Processing · doi:10.1109/icip.2018.8451419 · Cited by 7
“To enhance the quality of restoration, we want to use convolutional neural networks (CNNs) [19] because CNNs have achieved striking performance on signal restoration problems in 2-D images [20, 21, 22].”
MethodIterative Adaptive Nonconvex Low-Rank Tensor Approximation to Image Restoration Based on ADMM
Shen, Z., & Sun, H. (2018) · Journal of Mathematical Imaging and Vision · doi:10.1007/s10851-018-0867-0 · Cited by 7
“In paper [46], the deblurring and the denoising operations involved in image restoration are decoupled based on the Nash equilibrium; an iterative algorithmbased on decoupling of deblurring and denoising in the image restoration process is proposed in [47]; very recently, in order to take advantage of the merits of co…”
MethodVariational Inference for Computational Imaging Inverse Problems
Tonolini, F., Radford, J., Turpin, A., Faccio, D., & Murray‐Smith, R. (2019) · ENLIGHTEN (Jurnal Bimbingan dan Konseling Islam) · doi:10.48550/arxiv.1904.06264 · Cited by 23
Non‐uniform image blind deblurring by two‐stage fully convolution network
Wu, C., Wo, Y., Han, G., et al. (2020) · IET Image Processing · doi:10.1049/iet-ipr.2018.5716 · Cited by 6
A General Decoupled Learning Framework for Parameterized Image OperatorsPreprint
Fan, Q., Chen, D., Yuan, L., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.05852 · Cited by 7
“TABLE 9 Quantitative Results (PSNR) of the Image Denoising Task on the BSD68 Benchmark s BM3D [42] WNNM [43] TNRD [37] DCDP [44] Ours”
BackgroundLearning Filter Basis for Convolutional Neural Network CompressionPreprint
Li, Y., Gu, S., Gool, L. V., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.08932 · Cited by 7
“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-citationMulti-grained Attention Networks for Single Image Super-ResolutionPreprint
Wu, H., Zou, Z., Gui, J., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.11937 · Cited by 7
“We compare our MGAN with 8 state-of-the-art methods: SPMSR [48], SRCNN [7], FSRCNN [15], VDSR [8], IRCNN [49], SRMD [44], RDN [21], MSRN [20].”
MethodNon-Local ConvLSTM for Video Compression Artifact ReductionPreprint
Xu, Y., Gao, L., Tian, K., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1910.12286 · Cited by 7
“Afterwards, [22, 28, 35, 36, 49, 50] proposed deeper networks to further reduce compression artifacts.”
BackgroundBasis Prediction Networks for Effective Burst Denoising with Large KernelsPreprint
Xia, Z., Perazzi, F., Gharbi, M., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.04421 · Cited by 7
“Recent work has improved image denoising performance using convolutional networks trained on large datasets [3, 38, 39, 22, 40, 35, 36].”
Background
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
