Multi-channel Weighted Nuclear Norm Minimization for Real Color Image Denoising

By Jun Xu, Lei Zhang, David Dapeng Zhang, Xiangchu Feng. IEEE International Conference on Computer Vision, 2017

311 citations recorded by the source corpus.

Most of the existing denoising algorithms are developed for grayscale images. It is not trivial to extend them for color image denoising since the noise statistics in R, G, and B channels can be very different for real noisy images. In this paper, we propose a multi-channel (MC) optimization model for real color image denoising under the weighted nuclear norm minimization (WNNM) framework. We concatenate the RGB patches to make use of the channel redundancy, and introduce a weight matrix to balance the data fidelity of the three channels in consideration of their different noise statistics. The proposed MC-WNNM model does not have an analytical solution. We reformulate it into a linear equality-constrained problem and solve it via alternating direction method of multipliers. Each alternative updating step has a closed-form solution and the convergence can be guaranteed. Experiments on both synthetic and real noisy image datasets demonstrate the superiority of the proposed MC-WNNM over state-of-the-art denoising methods.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/dd897ce8c19663aa8f9811608f6ba0a344bf5d65

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Learning to Restore Low-Light Images via Decomposition-and-Enhancement (2020, methodology): “Many methods have been developed based on image priors such as selfsimilarity [3, 11], sparsity [13, 30], and low rank [18, 43].”
Human-Aware Motion Deblurring (2019, methodology): “Numerous CNN-based methods have also been applied in an endto-end fashion for image processing and generation problems, such as segmentation [24, 42], super-resolution [53, 22, 14, 15, 44], denoising [25, 49, 59, 47, 48], dehazing and deraining [32, 55], enhancement [31, 56] etc.”
Toward Convolutional Blind Denoising of Real Photographs (2018, methodology): “(e) NI [2] (f) MCWNNM [7] (g) TWSC [6] (h) CBDNet Figure 10: Denoising results of a SIDD image by different methods.”
High-Quality Self-Supervised Deep Image Denoising (2019, result): “3Even though (grayscale) WNNM [9] has been shown to be superior to (grayscale) BM3D [7], our experiments with the official implementation of MCWNNM [30], a multi-channel version of WNNM, indicated that CBM3D performs better on our test data where all color channels have the same amount of noise.”
Real Image Denoising With Feature Attention (2019, background): “31dB Noisy Image MCWNNM [58] NI [2] FFDNet [65] CBDNet [31] RIDNet (Ours) Figure 5.”
CycleISP: Real Image Restoration via Improved Data Synthesis (2020, methodology): “Method EPLL TNRD NCSR MLP BM3D FoE WNNM KSVD MCWNNM FFDNet+ TWSC CBDNet RIDNet Ours [69] [14] [18] [9] [15] [50] [27] [3] [62] [67] [61] [28] [4]”
Residual Dense Network for Image Restoration (2018, methodology): “Following the same settings as [74], we estimate the noise levels (σr, σg, σb) via some noise estimation methods [75].”
Unprocessing Images for Learned Raw Denoising (2018, unknown): “1%) ∼60,000 MCWNNM [40] - - 37.”
Spatial-Adaptive Network for Single Image Denoising (2020, unknown): “The common priors include local smoothing [23,30], sparsity [2,20,33], non-local self-similarity [5,9,8,34,11] and external statistical prior [41,32].”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, methodology): “Method EPLL TNRD MLP BM3D FoE WNNM KSVD MCWNNM FFDNet+ TWSC CBDNet RIDNet VDN MIRNet [128] [20] [11] [21] [85] [38] [2] [112] [121] [111] [39] [5] [118] (Ours)”
STAR: A Structure and Texture Aware Retinex Model (2019, methodology): “We solve our STAR model by an alternating optimization algorithm [49], [60].”