Unsupervised Image Restoration Using Partially Linear Denoisers

TL;DR

This paper introduces a novel unsupervised image restoration method using structured denoisers that can be trained solely on noisy images, eliminating the need for clean ground truth images and outperforming existing unsupervised models.

Contribution

It proposes a class of partially linear denoisers that are trainable with only noisy images, enabling effective image restoration without clean training data.

Findings

Outperforms recent unsupervised denoising models.

Effective for blind deblurring with only one noisy observation.

Achieves near-supervised quality on benchmark datasets.

Abstract

Deep neural network based methods are the state of the art in various image restoration problems. Standard supervised learning frameworks require a set of noisy measurement and clean image pairs for which a distance between the output of the restoration model and the ground truth, clean images is minimized. The ground truth images, however, are often unavailable or very expensive to acquire in real-world applications. We circumvent this problem by proposing a class of structured denoisers that can be decomposed as the sum of a nonlinear image-dependent mapping, a linear noise-dependent term and a small residual term. We show that these denoisers can be trained with only noisy images under the condition that the noise has zero mean and known variance. The exact distribution of the noise, however, is not assumed to be known. We show the superiority of our approach for image denoising, and demonstrate its extension to solving other restoration problems such as blind deblurring where the ground truth is not available. Our method outperforms some recent unsupervised and self-supervised deep denoising models that do not require clean images for their training. For blind deblurring problems, the method, using only one noisy and blurry observation per image, reaches a quality not far away from its fully supervised counterparts on a benchmark dataset.