When Schr\"odinger Bridge Meets Real-World Image Dehazing with Unpaired Training

TL;DR

DehazeSB introduces a Schr"odinger Bridge-based unpaired dehazing framework that leverages optimal transport theory and CLIP-based prompt learning to produce high-quality dehazed images efficiently.

Contribution

This work presents a novel unpaired dehazing method using Schr"odinger Bridge and optimal transport, enhancing transport mapping efficiency and integrating vision-language alignment.

Findings

Outperforms existing unpaired dehazing methods on real-world datasets.

Generates high-quality dehazed images with fewer steps.

Maintains structural details and consistency in restored images.

Abstract

Recent advancements in unpaired dehazing, particularly those using GANs, show promising performance in processing real-world hazy images. However, these methods tend to face limitations due to the generator's limited transport mapping capability, which hinders the full exploitation of their effectiveness in unpaired training paradigms. To address these challenges, we propose DehazeSB, a novel unpaired dehazing framework based on the Schr\"odinger Bridge. By leveraging optimal transport (OT) theory, DehazeSB directly bridges the distributions between hazy and clear images. This enables optimal transport mappings from hazy to clear images in fewer steps, thereby generating high-quality results. To ensure the consistency of structural information and details in the restored images, we introduce detail-preserving regularization, which enforces pixel-level alignment between hazy inputs and dehazed outputs. Furthermore, we propose a novel prompt learning to leverage pre-trained CLIP models in distinguishing hazy images and clear ones, by learning a haze-aware vision-language alignment. Extensive experiments on multiple real-world datasets demonstrate our method's superiority. Code: https://github.com/ywxjm/DehazeSB.