Noise2Noise: Learning Image Restoration without Clean Data

TL;DR

This paper introduces a method to train image restoration models using only corrupted images, demonstrating that clean signals can be learned without clean data or explicit priors, often matching or surpassing traditional methods.

Contribution

It shows that deep learning models can be trained solely on noisy data for various image restoration tasks, eliminating the need for clean training datasets.

Findings

Models trained on noisy data effectively remove photographic noise.

Successful denoising of synthetic Monte Carlo images.

Reconstruction of undersampled MRI scans using only corrupted data.

Abstract

We apply basic statistical reasoning to signal reconstruction by machine learning -- learning to map corrupted observations to clean signals -- with a simple and powerful conclusion: it is possible to learn to restore images by only looking at corrupted examples, at performance at and sometimes exceeding training using clean data, without explicit image priors or likelihood models of the corruption. In practice, we show that a single model learns photographic noise removal, denoising synthetic Monte Carlo images, and reconstruction of undersampled MRI scans -- all corrupted by different processes -- based on noisy data only.