Exploiting Diffusion Prior for Real-World Image Dehazing with Unpaired Training

2025-03-19

AI Summary

Addressing the challenging generalization problem of single-image dehazing under real-world conditions—where paired hazy and haze-free images are unavailable—this paper proposes Diff-Dehazer, the first unpaired dehazing framework integrating diffusion priors. Methodologically, it innovatively embeds a pre-trained diffusion model as a bidirectional mapping learner into CycleGAN, jointly leveraging physical imaging priors and a cross-modal (image + text) degradation modeling mechanism to explicitly capture real-world haze distributions and semantic knowledge. Extensive experiments on multiple real-world datasets demonstrate that Diff-Dehazer significantly outperforms existing state-of-the-art methods, with both qualitative and quantitative results confirming its superior generalization capability and robustness across diverse haze conditions. The source code is publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• 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

Abstract

Unpaired training has been verified as one of the most effective paradigms for real scene dehazing by learning from unpaired real-world hazy and clear images. Although numerous studies have been proposed, current methods demonstrate limited generalization for various real scenes due to limited feature representation and insufficient use of real-world prior. Inspired by the strong generative capabilities of diffusion models in producing both hazy and clear images, we exploit diffusion prior for real-world image dehazing, and propose an unpaired framework named Diff-Dehazer. Specifically, we leverage diffusion prior as bijective mapping learners within the CycleGAN, a classic unpaired learning framework. Considering that physical priors contain pivotal statistics information of real-world data, we further excavate real-world knowledge by integrating physical priors into our framework. Furthermore, we introduce a new perspective for adequately leveraging the representation ability of diffusion models by removing degradation in image and text modalities, so as to improve the dehazing effect. Extensive experiments on multiple real-world datasets demonstrate the superior performance of our method. Our code https://github.com/ywxjm/Diff-Dehazer.

Problem

• Addresses limited generalization in real-world image dehazing.
• Exploits diffusion models for better feature representation.
• Integrates physical priors to enhance dehazing performance.

Innovation

• Leverages diffusion prior for image dehazing
• Integrates physical priors into CycleGAN framework
• Removes degradation in image and text modalities