Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising

@article{Zhang2016BeyondAG,
  title={Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising},
  author={K. Zhang and Wangmeng Zuo and Yunjin Chen and Deyu Meng and Lei Zhang},
  journal={IEEE Transactions on Image Processing},
  year={2016},
  volume={26},
  pages={3142-3155}
}
  • K. ZhangW. Zuo Lei Zhang
  • Published 13 August 2016
  • Computer Science
  • IEEE Transactions on Image Processing
The discriminative model learning for image denoising has been recently attracting considerable attentions due to its favorable denoising performance. [] Key Method With the residual learning strategy, DnCNN implicitly removes the latent clean image in the hidden layers. This property motivates us to train a single DnCNN model to tackle with several general image denoising tasks, such as Gaussian denoising, single image super-resolution, and JPEG image deblocking. Our extensive experiments demonstrate that…

A Simple and Robust Deep Convolutional Approach to Blind Image Denoising

Experimental results demonstrate that the proposed blind denoiser achieves fairly competitive or even better performance than state-of-the-art algorithms in terms of both quantitative and qualitative assessment.

Dilated Deep Residual Network for Image Denoising

A dilated residual CNN for Gaussian image denoising by adopting dilated convolution in residual network, and the dilation factor is set to a certain value so that the change of receptive field can be interpreted mathematically.

An adaptive image denoising method based on Deep Rectified Denoising Auto-Encoder

An adaptive image denoising method based on Deep Rectified Denoising Auto-Encoder that can not only outperform SSDA in the convergence speed, but also adaptively remove the noise that is not seen in training.

Dilated Residual Network for Image Denoising

A dilated residual CNN for Gaussian image denoising by adopting dilated convolution in residual network, and the dilation factor is set to a certain value so that the change of receptive field can be interpreted mathematically.

Variational Denoising Network: Toward Blind Noise Modeling and Removal

This work proposes a new variational inference method, which integrates both noise estimation and image denoising into a unique Bayesian framework, for blind image Denoising, and presents an approximate posterior, parameterized by deep neural networks, presented by taking the intrinsic clean image and noise variances as latent variables conditioned on the input noisy image.

Deep Learning on Image Denoising: An overview

Few-Shot Meta-Learning for Image Denoising

The main objective of the Meta-Denoiser is to find an ideal set of weights to initialize a denoising convolutional neural network (DnCNN), so that a DnCNN can adapt to unseen noise distributions with very few training examples.

Image Denoising using Attention-Residual Convolutional Neural Networks

The proposed methods try to learn the underlying noise expectation using an Attention-Residual mechanism, and demonstrate the effectiveness of the proposed approaches against some state-of-the-art image denoising methods.

Self-supervised Denoising via Low-rank Tensor Approximated Convolutional Neural Network

A new self-supervised framework for image denoising based on the Tucker low-rank tensor approximation is introduced, able to characterize the authors' denoiser with fewer parameters and train it based on a single image, which considerably improves the model generalizability and reduces the cost of data acquisition.

Boosting of Denoising Effect with Fusion Strategy

The proposed method significantly outperforms the NCSR and the DnCNN both quantitatively and visually when they are considered as individual methods; similarly, it outperforms several other state-of-the-art denoising methods.
...

References

SHOWING 1-10 OF 49 REFERENCES

Image Denoising and Inpainting with Deep Neural Networks

A novel approach to low-level vision problems that combines sparse coding and deep networks pre-trained with denoising auto-encoder (DA) is presented and can automatically remove complex patterns like superimposed text from an image, rather than simple patterns like pixels missing at random.

Natural Image Denoising with Convolutional Networks

An approach to low-level vision is presented that combines the use of convolutional networks as an image processing architecture and an unsupervised learning procedure that synthesizes training samples from specific noise models to avoid computational difficulties in MRF approaches that arise from probabilistic learning and inference.

Image Denoising Via Sparse and Redundant Representations Over Learned Dictionaries

This work addresses the image denoising problem, where zero-mean white and homogeneous Gaussian additive noise is to be removed from a given image, and uses the K-SVD algorithm to obtain a dictionary that describes the image content effectively.

Patch Group Based Nonlocal Self-Similarity Prior Learning for Image Denoising

It is demonstrated that, owe to the learned PG-GMM, a simple weighted sparse coding model, which has a closed-form solution, can be used to perform image denoising effectively, resulting in high PSNR measure, fast speed, and particularly the best visual quality among all competing methods.

Cascades of Regression Tree Fields for Image Restoration

A cascade model for image restoration that consists of a Gaussian CRF at each stage that is semi-parametric, i.e., it depends on the instance-specific parameters of the restoration problem, such as the blur kernel.

Learning real-time MRF inference for image denoising

  • Adrian Barbu
  • Computer Science
    2009 IEEE Conference on Computer Vision and Pattern Recognition
  • 2009
This paper argues that through appropriate training, a MRF/CRF model can be trained to perform very well on a suboptimal inference algorithm, and performs close to state-of-the-art accuracy in image denoising.

Natural image denoising: Optimality and inherent bounds

This paper takes a non parametric approach and represents the distribution of natural images using a huge set of 1010 patches and derives a simple statistical measure which provides a lower bound on the optimal Bayesian minimum mean square error (MMSE).

Non-local sparse models for image restoration

Experimental results in image denoising and demosaicking tasks with synthetic and real noise show that the proposed method outperforms the state of the art, making it possible to effectively restore raw images from digital cameras at a reasonable speed and memory cost.

Deep Residual Learning for Image Recognition

This work presents a residual learning framework to ease the training of networks that are substantially deeper than those used previously, and provides comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth.

Patch Complexity, Finite Pixel Correlations and Optimal Denoising

A law of diminishing return is presented, namely that with increasing patch size, rare patches not only require a much larger dataset, but also gain little from it, and this result suggests novel adaptive variable-sized patch schemes for denoising.