When Schrödinger Bridge Meets Real-World Image Dehazing with Unpaired Training
This work addresses image quality enhancement for computer vision applications, offering a novel method for unpaired dehazing with potential improvements over existing GAN-based approaches.
The paper tackles the problem of real-world image dehazing with unpaired training by proposing DehazeSB, a framework based on the Schrödinger Bridge that bridges distributions between hazy and clear images using optimal transport theory, resulting in high-quality dehazed images as demonstrated by extensive experiments on multiple datasets.
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 Schrödinger 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.