Noise2Void - Learning Denoising From Single Noisy Images

By Alexander Krull, Tim-Oliver Buchholz, Florian Jug. Computer Vision and Pattern Recognition, 2018

1,490 citations recorded by the source corpus.

The field of image denoising is currently dominated by discriminative deep learning methods that are trained on pairs of noisy input and clean target images. Recently it has been shown that such methods can also be trained without clean targets. Instead, independent pairs of noisy images can be used, in an approach known as Noise2Noise (N2N). Here, we introduce Noise2Void (N2V), a training scheme that takes this idea one step further. It does not require noisy image pairs, nor clean target images. Consequently, N2V allows us to train directly on the body of data to be denoised and can therefore be applied when other methods cannot. Especially interesting is the application to biomedical image data, where the acquisition of training targets, clean or noisy, is frequently not possible. We compare the performance of N2V to approaches that have either clean target images and/or noisy image pairs available. Intuitively, N2V cannot be expected to outperform methods that have more information available during training. Still, we observe that the denoising performance of Noise2Void drops in moderation and compares favorably to training-free denoising methods.

Subfields: Biophysics, Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/be2cafced16bb8834bdd322a0a512142c8d05388

How researchers see this paper

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

In Biophysics

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

NeRF in the Dark: High Dynamic Range View Synthesis from Noisy Raw Images (2021, background): “Various followups to Noise2Noise proposed modiﬁed network architectures allowing supervision with a dataset of single noisy images [4, 28, 29].”
Democratising deep learning for microscopy with ZeroCostDL4Mic (2021, methodology): “We ﬁ rst demonstrated the capabilities of our Noise2Void notebook to denoise the movie of an ovarian carcinoma cell migrating on cell-derived matrices (Supplementary Movie 7).”
Self-Supervised Learning of Graph Neural Networks: A Unified Review (2021, background): “Recently, SSL has shown its promising capability in data restoration tasks, such as image super-resolution [1], image denoising [2, 3, 4], and single-cell analysis [5].”
Cellpose3: one-click image restoration for improved cellular segmentation (2024, neutral): “We compared the performance of the Cellpose3 models to Noise2Void, Noise2Self, and CARE [7, 14, 15].”
Learning to Restore Low-Light Images via Decomposition-and-Enhancement (2020, methodology): “images, by synthesizing noise in the raw data domain [2], constructing real image dataset [1], developing joint training strategy of both synthetic and real images [19], or unsupervised learning [28].”
Noisier2Noise: Learning to Denoise From Unpaired Noisy Data (2019, background): “Both [8] and [1] propose to learn to perform denoising using only a single noisy realization of each training sample.”
Artificial intelligence in the creative industries: a review (2020, methodology): “n problem separately for each input sample. However, this algorithm is sensitive to the loss function used, which can signicantly in uence the performance of the model. Another algorithm, Noise2Void [130], employs a novel blind-spot network that is not include the current pixel in the convolution. T”
Self2Self With Dropout: Learning Self-Supervised Denoising From Single Image (2020, methodology): “Following N2V’s setting and its noisy data generation scheme, we train N2V and N2S on CBSD300 [15] and CBSD300’s gray-scale version for color/gray-scale image denoising respectively.”
High-Quality Self-Supervised Deep Image Denoising (2019, methodology): “[14] so that a blind spot is always intentional.”
Neighbor2Neighbor: Self-Supervised Denoising from Single Noisy Images (2021, methodology): “Recently, Probabilistic Noise2Void [14], Laine19 [15], and Dilated Blind-Spot Network [30] further introduced explicit noise modeling and probabilistic inference for better performance.”
BockNet: Blind-Block Reconstruction Network With a Guard Window for Hyperspectral Anomaly Detection (2023, background): “Consequently, the network’s prediction for a pixel is influenced by all surrounding pixels within the receptive field, excluding the central one [65].”
You Only Look Yourself: Unsupervised and Untrained Single Image Dehazing Neural Network (2020, methodology): “…hazy image, DehazeNet (Cai et al., 2016), MSCNN (Ren et al., 2016), AOD-Net (Li et al., 2017), DCP (He et al., 2009), GRM (Chen et al., 2016), N2V (Krull et al., 2019), DIP (Ulyanov et al., 2018), DD (Heckel and Hand, 2019), DDIP (Irani, 2019), our YOLY and the ground truth are presented in…”