Noise2Score: Tweedie's Approach to Self-Supervised Image Denoising without Clean Images

TL;DR

Noise2Score introduces a unified self-supervised image denoising framework based on Tweedie's formula, capable of handling various noise types without clean images, and significantly outperforms existing methods in benchmark tests.

Contribution

The paper presents a novel method that unifies existing self-supervised denoising approaches using Tweedie's formula and score function estimation, enabling versatile noise removal.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Works effectively with Gaussian, Poisson, and Gamma noises

Uses a single network for multiple noise types

Abstract

Recently, there has been extensive research interest in training deep networks to denoise images without clean reference. However, the representative approaches such as Noise2Noise, Noise2Void, Stein's unbiased risk estimator (SURE), etc. seem to differ from one another and it is difficult to find the coherent mathematical structure. To address this, here we present a novel approach, called Noise2Score, which reveals a missing link in order to unite these seemingly different approaches. Specifically, we show that image denoising problems without clean images can be addressed by finding the mode of the posterior distribution and that the Tweedie's formula offers an explicit solution through the score function (i.e. the gradient of log likelihood). Our method then uses the recent finding that the score function can be stably estimated from the noisy images using the amortized residual denoising autoencoder, the method of which is closely related to Noise2Noise or Nose2Void. Our Noise2Score approach is so universal that the same network training can be used to remove noises from images that are corrupted by any exponential family distributions and noise parameters. Using extensive experiments with Gaussian, Poisson, and Gamma noises, we show that Noise2Score significantly outperforms the state-of-the-art self-supervised denoising methods in the benchmark data set such as (C)BSD68, Set12, and Kodak, etc.