UniRes: Universal Image Restoration for Complex Degradations

2025-06-05

AI Summary

Real-world image restoration faces challenges from complex, mixed unknown degradations, and existing methods suffer from limited generalization. This paper proposes UniRes, an end-to-end diffusion framework that introduces, for the first time, a multi-task knowledge transfer–driven diffusion sampling paradigm: during single-step denoising, it dynamically fuses knowledge from multiple specialized restoration models via degradation-aware conditional injection and differentiable weight scheduling, enabling adaptive collaborative restoration. UniRes requires only independently collected training data per degradation type, supports flexible model extension, and allows tunable fidelity–quality trade-offs. On benchmarks with complex degradations, it surpasses state-of-the-art methods by 1.2–2.8 dB in PSNR and SSIM, while maintaining competitive performance on single-degradation tasks—significantly enhancing model generalization and robustness.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Real-world image restoration is hampered by diverse degradations stemming from varying capture conditions, capture devices and post-processing pipelines. Existing works make improvements through simulating those degradations and leveraging image generative priors, however generalization to in-the-wild data remains an unresolved problem. In this paper, we focus on complex degradations, i.e., arbitrary mixtures of multiple types of known degradations, which is frequently seen in the wild. A simple yet flexible diffusionbased framework, named UniRes, is proposed to address such degradations in an end-to-end manner. It combines several specialized models during the diffusion sampling steps, hence transferring the knowledge from several well-isolated restoration tasks to the restoration of complex in-the-wild degradations. This only requires well-isolated training data for several degradation types. The framework is flexible as extensions can be added through a unified formulation, and the fidelity-quality trade-off can be adjusted through a new paradigm. Our proposed method is evaluated on both complex-degradation and single-degradation image restoration datasets. Extensive qualitative and quantitative experimental results show consistent performance gain especially for images with complex degradations.

Problem

• Addresses complex image degradations from diverse real-world sources
• Improves generalization to in-the-wild data with mixed degradations
• Proposes flexible diffusion-based framework for universal restoration

Innovation

• Diffusion-based framework for complex degradations
• Combines specialized models in sampling steps
• Flexible extensions through unified formulation