Noise2Same: Optimizing A Self-Supervised Bound for Image Denoising

TL;DR

Noise2Same introduces a new self-supervised image denoising framework that does not rely on J-invariance, achieving superior performance and efficiency by deriving a novel loss bound, applicable to a wider range of noise models.

Contribution

We propose Noise2Same, a self-supervised denoising method that removes the need for J-invariance and noise model assumptions, with a new loss derived from an upper bound of supervised loss.

Findings

Outperforms previous self-supervised denoising methods in accuracy.

Requires no J-invariance or noise model information.

Improves training efficiency and denoising performance.

Abstract

Self-supervised frameworks that learn denoising models with merely individual noisy images have shown strong capability and promising performance in various image denoising tasks. Existing self-supervised denoising frameworks are mostly built upon the same theoretical foundation, where the denoising models are required to be J-invariant. However, our analyses indicate that the current theory and the J-invariance may lead to denoising models with reduced performance. In this work, we introduce Noise2Same, a novel self-supervised denoising framework. In Noise2Same, a new self-supervised loss is proposed by deriving a self-supervised upper bound of the typical supervised loss. In particular, Noise2Same requires neither J-invariance nor extra information about the noise model and can be used in a wider range of denoising applications. We analyze our proposed Noise2Same both theoretically and experimentally. The experimental results show that our Noise2Same remarkably outperforms previous self-supervised denoising methods in terms of denoising performance and training efficiency. Our code is available at https://github.com/divelab/Noise2Same.