Published on Fri Jan 08 2021

Neighbor2Neighbor: Self-Supervised Denoising from Single Noisy Images

Tao Huang, Songjiang Li, Xu Jia, Huchuan Lu, Jianzhuang Liu

Neighbor2Neighbor is a method to train an effective image denoising model with only noisy images. A random neighbor sub-sampler is proposed for the generation of training image pairs. A regularizer is proposed as additional loss for better performance.

0
0
0
Abstract

In the last few years, image denoising has benefited a lot from the fast development of neural networks. However, the requirement of large amounts of noisy-clean image pairs for supervision limits the wide use of these models. Although there have been a few attempts in training an image denoising model with only single noisy images, existing self-supervised denoising approaches suffer from inefficient network training, loss of useful information, or dependence on noise modeling. In this paper, we present a very simple yet effective method named Neighbor2Neighbor to train an effective image denoising model with only noisy images. Firstly, a random neighbor sub-sampler is proposed for the generation of training image pairs. In detail, input and target used to train a network are images sub-sampled from the same noisy image, satisfying the requirement that paired pixels of paired images are neighbors and have very similar appearance with each other. Secondly, a denoising network is trained on sub-sampled training pairs generated in the first stage, with a proposed regularizer as additional loss for better performance. The proposed Neighbor2Neighbor framework is able to enjoy the progress of state-of-the-art supervised denoising networks in network architecture design. Moreover, it avoids heavy dependence on the assumption of the noise distribution. We explain our approach from a theoretical perspective and further validate it through extensive experiments, including synthetic experiments with different noise distributions in sRGB space and real-world experiments on a denoising benchmark dataset in raw-RGB space.

Mon Jun 17 2019
Computer Vision
Noisy-As-Clean: Learning Self-supervised Denoising from the Corrupted Image
Supervised deep networks have achieved promisingperformance on image denoising. However, for an unseen corrupted image, both supervised and unsupervised networks ignore either its particular image prior, the noise statistics, or both. To circumvent this problem, we propose a novel "Noisy-As-Clean" strategy.
0
0
0
Fri Nov 06 2020
Machine Learning
Noise2Sim -- Similarity-based Self-Learning for Image Denoising
Noise2Noise is an NLM-inspired self-learning method for image denoising. Noise2Sim leverages the self-similarity of image pixels to train the network, requiring single noisy images only. Extensive experiments demonstrate the superiority of Noise 2Sim on common benchmark datasets.
3
0
0
Thu Oct 22 2020
Machine Learning
Noise2Same: Optimizing A Self-Supervised Bound for Image Denoising
Noise2Same is a novel self-supervised denoising framework. Noise2Same requires neither J-invariance nor extra information about the noise model. It can be used in a wider range of denoised applications.
0
0
0
Tue Dec 03 2019
Computer Vision
Noise2Blur: Online Noise Extraction and Denoising
Noise2Blur (N2B) is a framework for training robust image Denoising models without pre-collected paired noisy/clean images. The model consists of two parts: a denoising network and a noise extraction network.
0
0
0
Mon Mar 09 2020
Computer Vision
Restore from Restored: Single Image Denoising with Pseudo Clean Image
Many supervised denoising approaches can produce satisfactory results using large external training datasets. However, these methods have limitations in using internal information available in a given test image. We propose a simple and effective fine-tuning algorithm called "restore-from-restored"
0
0
0
Mon Aug 31 2020
Computer Vision
Unpaired Learning of Deep Image Denoising
We investigate the task of learning blind image denoising networks from an unpaired set of clean and noisy images. We assume that the noise can be signal dependent but is spatially uncorrelated. We first apply the learned noise models to clean images to synthesize a paired set of training images.
0
0
0