IRBridge: Solving Image Restoration Bridge with Pre-trained Generative Diffusion Models

2025-05-30

AI Summary

Existing bridging models for image restoration require separate training for each degradation type, resulting in high computational costs and poor generalization. Method: This paper proposes a fine-tuning-free zero-shot transfer paradigm. Its core innovation is a unified transition equation that jointly parameterizes both the noise起点 and degradation起点, enabling plug-and-play adaptation of pre-trained generative diffusion models (e.g., Stable Diffusion) to restoration tasks. By aligning state distributions within a diffusion bridging framework, the method bridges the mismatch between generative priors and degradation distributions. Contribution/Results: The approach achieves cross-degradation generalization without task-specific training. It is validated across six diverse image restoration tasks, demonstrating strong robustness and generalization. Crucially, it enables training-free deployment and reduces computational overhead by over 90% compared to conventional methods.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Bridge models in image restoration construct a diffusion process from degraded to clear images. However, existing methods typically require training a bridge model from scratch for each specific type of degradation, resulting in high computational costs and limited performance. This work aims to efficiently leverage pretrained generative priors within existing image restoration bridges to eliminate this requirement. The main challenge is that standard generative models are typically designed for a diffusion process that starts from pure noise, while restoration tasks begin with a low-quality image, resulting in a mismatch in the state distributions between the two processes. To address this challenge, we propose a transition equation that bridges two diffusion processes with the same endpoint distribution. Based on this, we introduce the IRBridge framework, which enables the direct utilization of generative models within image restoration bridges, offering a more flexible and adaptable approach to image restoration. Extensive experiments on six image restoration tasks demonstrate that IRBridge efficiently integrates generative priors, resulting in improved robustness and generalization performance. Code will be available at GitHub.

Problem

• Leveraging pretrained generative models for image restoration bridges
• Addressing mismatch between generative and restoration diffusion processes
• Enhancing robustness and generalization in diverse restoration tasks

Innovation

• Bridges diffusion processes with transition equation
• Utilizes pretrained generative models directly
• Enhances robustness and generalization performance