Towards Robust Image Denoising with Scale Equivariance

TL;DR

This paper introduces a scale-equivariant approach for image denoising that enhances robustness to out-of-distribution, spatially non-uniform noise by incorporating novel modules that stabilize features and modulate information.

Contribution

We propose a new scale-equivariant denoising framework with Heterogeneous Normalization and Interactive Gating modules to improve OOD robustness in image denoising.

Findings

Outperforms state-of-the-art methods on synthetic benchmarks

Achieves superior results on real-world noisy images

Demonstrates robustness to spatially heterogeneous noise

Abstract

Despite notable advances in image denoising, existing models often struggle to generalize beyond in-distribution noise patterns, particularly when confronted with out-of-distribution (OOD) conditions characterized by spatially variant noise. This generalization gap remains a fundamental yet underexplored challenge. In this work, we investigate \emph{scale equivariance} as a core inductive bias for improving OOD robustness. We argue that incorporating scale-equivariant structures enables models to better adapt from training on spatially uniform noise to inference on spatially non-uniform degradations. Building on this insight, we propose a robust blind denoising framework equipped with two key components: a Heterogeneous Normalization Module (HNM) and an Interactive Gating Module (IGM). HNM stabilizes feature distributions and dynamically corrects features under varying noise intensities, while IGM facilitates effective information modulation via gated interactions between signal and feature paths. Extensive evaluations demonstrate that our model consistently outperforms state-of-the-art methods on both synthetic and real-world benchmarks, especially under spatially heterogeneous noise. Code will be made publicly available.