ReviveDiff: A Universal Diffusion Model for Restoring Images in Adverse Weather Conditions

2024-09-27

arXiv.org

AI Summary

Adverse weather conditions—including nighttime, rain, fog, smoke, underwater scenes, and low illumination—induce severe image degradation; existing methods typically target only a single degradation type, suffering from poor generalizability. This paper proposes ReviveDiff, the first diffusion-based universal image restoration model tailored to diverse natural-medium degradations. ReviveDiff pioneers the application of conditional diffusion models to unified cross-domain adverse-weather image restoration. It introduces a novel macro–micro collaborative optimization framework that jointly integrates multi-scale feature reconstruction, physics-informed degradation modeling, and multi-task quality-factor optimization—simultaneously enhancing sharpness, dynamic range, color fidelity, and structural preservation. Evaluated across seven benchmark datasets spanning five degradation categories, ReviveDiff consistently outperforms state-of-the-art methods, achieving average improvements of 1.8 dB in PSNR and 0.032 in SSIM, with visually more natural and artifact-free results.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Adversarial Learning & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Images captured in challenging environments--such as nighttime, smoke, rainy weather, and underwater--often suffer from significant degradation, resulting in a substantial loss of visual quality. The effective restoration of these degraded images is critical for the subsequent vision tasks. While many existing approaches have successfully incorporated specific priors for individual tasks, these tailored solutions limit their applicability to other degradations. In this work, we propose a universal network architecture, dubbed ``ReviveDiff'', which can address various degradations and bring images back to life by enhancing and restoring their quality. Our approach is inspired by the observation that, unlike degradation caused by movement or electronic issues, quality degradation under adverse conditions primarily stems from natural media (such as fog, water, and low luminance), which generally preserves the original structures of objects. To restore the quality of such images, we leveraged the latest advancements in diffusion models and developed ReviveDiff to restore image quality from both macro and micro levels across some key factors determining image quality, such as sharpness, distortion, noise level, dynamic range, and color accuracy. We rigorously evaluated ReviveDiff on seven benchmark datasets covering five types of degrading conditions: Rainy, Underwater, Low-light, Smoke, and Nighttime Hazy. Our experimental results demonstrate that ReviveDiff outperforms the state-of-the-art methods both quantitatively and visually.

Problem

• Restoring images degraded by adverse weather conditions
• Developing a universal model for multiple degradation types
• Enhancing image quality factors like sharpness and color accuracy

Innovation

• Universal diffusion model for image restoration
• Macro and micro level quality enhancement
• Handles multiple adverse weather conditions