Self-Calibrated Variance-Stabilizing Transformations for Real-World Image Denoising

TL;DR

This paper introduces Noise2VST, a method that applies a variance-stabilizing transform to real-world noisy images, enabling effective denoising with Gaussian-trained networks without additional training data.

Contribution

The paper proposes Noise2VST, a novel algorithm for learning a model-free variance-stabilizing transform that improves real-world image denoising using existing Gaussian denoisers.

Findings

Noise2VST outperforms existing methods in real-world denoising tasks.

The approach requires only the noisy image and a standard Gaussian denoiser.

Extensive experiments demonstrate the method's efficiency and superiority.

Abstract

Supervised deep learning has become the method of choice for image denoising. It involves the training of neural networks on large datasets composed of pairs of noisy and clean images. However, the necessity of training data that are specific to the targeted application constrains the widespread use of denoising networks. Recently, several approaches have been developed to overcome this difficulty by whether artificially generating realistic clean/noisy image pairs, or training exclusively on noisy images. In this paper, we show that, contrary to popular belief, denoising networks specialized in the removal of Gaussian noise can be efficiently leveraged in favor of real-world image denoising, even without additional training. For this to happen, an appropriate variance-stabilizing transform (VST) has to be applied beforehand. We propose an algorithm termed Noise2VST for the learning of such a model-free VST. Our approach requires only the input noisy image and an off-the-shelf Gaussian denoiser. We demonstrate through extensive experiments the efficiency and superiority of Noise2VST in comparison to existing methods trained in the absence of specific clean/noisy pairs.