Image denoising: Can plain neural networks compete with BM3D?

By Harold Christopher Burger, Christian J. Schuler, Stefan Harmeling. 2012

1,345 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

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

Non-local Neural Networks (2017, background): “BM3D is a solid image denoising baseline even compared with deep neural networks [5].”
N4-Fields: Neural Network Nearest Neighbor Fields for Image Transforms (2014, background): “Another series of works [14, 15] investigate the use of convolutional neural networks for image denoising.”
Solving Linear Inverse Problems Using the Prior Implicit in a Denoiser (2020, background): “One expresses the estimation function (as opposed to the prior) in parametric form, and sets the parameters by minimizing the denoising MSE over a large training set of example signals and their noise-corrupted counterparts [23–26].”
Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising (2016, methodology): “MLP [24] ﬁrst uses a patch of size 39 × 39 to generate the predicted patch, and then adopts a ﬁlter of size 9 × 9 to average the output patches, thus its effective patch size is 47 × 47 .”
Restoring an Image Taken through a Window Covered with Dirt or Rain (2013, methodology): “When trained at the patch level, as in the system described by [2], each prediction leaves the same residual trace of the noise, which their average then maintains (b).”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, methodology): “Method DnCNN MLP GLIDE TNRD FoE BM3D WNNM NLM KSVD EPLL CBDNet RIDNet VDN MIRNet [120] [11] [96] [20] [85] [21] [38] [10] [2] [128] [39] [5] [118] (Ours)”
Computer Vision with a Single (Robust) Classifier (2019, background): “Super-resolution refers to the task of recovering high-resolution images given their low resolution version [DFE07; BSH12].”
Non-Local Recurrent Network for Image Restoration (2018, methodology): “The pioneering works include a multilayer perceptron for image denoising [3] and a three-layer CNN for image SR [10].”
A Statistical Prediction Model Based on Sparse Representations for Single Image Super-Resolution (2014, background): “Recently, a neural network has been suggested for the problem of image denoising [37], showing excellent performance, even though deep learning was not utilized there.”
Using Deep Neural Networks for Inverse Problems in Imaging: Beyond Analytical Methods (2018, methodology): “[8] (2013) Nonblind image deconvolution MLP with two hidden layers; trained to remove artifacts following nonblind deconvolution step”
Video Frame Interpolation via Adaptive Convolution (2017, background): “Our work is inspired by the success of deep learning in solving not only difficult visual understanding problems [16, 20, 26, 28, 39, 40, 42, 45, 54, 60, 64] but also other computer vision problems like optical flow estimation [9, 14, 19, 48, 49, 52], style transfer [11, 15, 23, 30, 50], and image e”