Degradation-Aware and Structure-Preserving Diffusion for Real-World Image Super-Resolution

TL;DR

This paper introduces a novel diffusion-based super-resolution method that explicitly models degradation and preserves structure, resulting in more realistic and high-quality image restoration in real-world scenarios.

Contribution

It proposes degradation-aware token injection and spatially asymmetric noise injection modules that improve real-world image super-resolution by explicitly handling complex degradations and structural details.

Findings

Achieves competitive perceptual quality on DIV2K and RealSR datasets.

Outperforms recent baselines in visual realism and perception-distortion trade-off.

Modules are lightweight and easily integrated into existing diffusion SR frameworks.

Abstract

Real-world image super-resolution is particularly challenging for diffusion models because real degradations are complex, heterogeneous, and rarely modeled explicitly. We propose a degradation-aware and structure-preserving diffusion framework for real-world SR. Specifically, we introduce Degradation-aware Token Injection, which encodes lightweight degradation statistics from low-resolution inputs and fuses them with semantic conditioning features, enabling explicit degradation-aware restoration. We further propose Spatially Asymmetric Noise Injection, which modulates diffusion noise with local edge strength to better preserve structural regions during training. Both modules are lightweight add-ons to the adopted diffusion SR framework, requiring only minor modifications to the conditioning pipeline. Experiments on DIV2K and RealSR show that our method delivers competitive no-reference perceptual quality and visually more realistic restoration results than recent baselines, while maintaining a favorable perception--distortion trade-off. Ablations confirm the effectiveness of each module and their complementary gains when combined. The code and model are publicly available at https://github.com/jiyang0315/DASP-SR.git.