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

2024-07-24

arXiv.org

AI Summary

Real-world image denoising suffers from the absence of paired noisy-clean data, making it challenging to leverage pre-trained Gaussian denoisers. To address this, we propose Noise2VST: a model-agnostic, self-supervised variance-stabilizing transform (VST) learning algorithm that requires no additional training. Given only a single real noisy image and an off-the-shelf Gaussian denoiser, Noise2VST iteratively self-calibrates by jointly exploiting statistical VST modeling and deep image priors. We provide the first theoretical proof—and empirical validation—that Gaussian denoisers can generalize directly to real noise distributions. Extensive experiments on multiple real-noise benchmarks demonstrate that Noise2VST significantly outperforms existing unpaired denoising methods, achieving state-of-the-art performance in both PSNR and perceptual quality.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Unsupervised & Self-Supervised Learning
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Security and Privacy: Privacy-enhancing technologies

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.

Problem

• Overcoming reliance on paired noisy-clean training data for denoising
• Adapting Gaussian noise denoisers for real-world image denoising
• Learning model-free variance-stabilizing transforms without clean references

Innovation

• Uses self-calibrated variance-stabilizing transformations
• Leverages Gaussian denoisers without retraining
• Learns model-free VST via Noise2VST algorithm