Publication detail

Probabilistic Noise2Void: Unsupervised Content-Aware Denoising

KRULL, A. VIČAR, T. PRAKASH, M. LALIT, M. JUG, F.

Original Title

Probabilistic Noise2Void: Unsupervised Content-Aware Denoising

Type

journal article in Web of Science

Language

English

Original Abstract

Today, Convolutional Neural Networks (CNNs) are the leading method for image denoising. They are traditionally trained on pairs of images, which are often hard to obtain for practical applications. This motivates self-supervised training methods, such as Noise2Void (N2V) that operate on single noisy images. Self-supervised methods are, unfortunately, not competitive with models trained on image pairs. Here, we present Probabilistic Noise2Void (PN2V), a method to train CNNs to predict per-pixel intensity distributions. Combining these with a suitable description of the noise, we obtain a complete probabilistic model for the noisy observations and true signal in every pixel. We evaluate PN2V on publicly available microscopy datasets, under a broad range of noise regimes, and achieve competitive results with respect to supervised state-of-the-art methods.

Keywords

denoising, CARE, deep learning, microscopy data, probabilistic

Authors

KRULL, A.; VIČAR, T.; PRAKASH, M.; LALIT, M.; JUG, F.

Released

19. 2. 2020

Publisher

Frontiers Media SA

ISBN

2624-9898

Periodical

Frontiers in Computer Science

Year of study

2

Number

5

State

Swiss Confederation

Pages from

1

Pages to

9

Pages count

9

URL

Full text in the Digital Library

BibTex

@article{BUT159778,
  author="Alexander {Krull} and Tomáš {Vičar} and Mangal {Prakash} and Manan {Lalit} and Florian {Jug}",
  title="Probabilistic Noise2Void: Unsupervised Content-Aware Denoising",
  journal="Frontiers in Computer Science",
  year="2020",
  volume="2",
  number="5",
  pages="1--9",
  doi="10.3389/fcomp.2020.00005",
  issn="2624-9898",
  url="https://www.frontiersin.org/articles/10.3389/fcomp.2020.00005/full"
}