Unleashing the Power of Deep Dehazing Models: A Physics-guided Parametric Augmentation Net for Image Rehazing
Keywords: Image dehazing, Image rehazing, Data augmentation
Abstract: Image dehazing faces significant challenges in real-world scenarios due to the large domain gap between synthetic and real-world hazy images, which often hinders dehazing performance. Collecting real-world datasets is particularly difficult, as hazy and clean image pairs must be captured under identical conditions. To address this, we propose a Physics-guided Parametric Augmentation Network (PANet) that generates realistic hazy and clean training pairs, enhancing dehazing performance in real-world applications. PANet consists of two components: a Haze-to-Parameter Mapper (HPM), which projects hazy images into a parametric space representing haze characteristics, and a Parameter-to-Haze Mapper (PHM), which converts resampled haze parameters back into hazy images. By resampling individual haze parameter maps at the pixel level in the parametric space, PANet generates diverse hazy images with physically explainable haze conditions that are not present in the training data. Our experimental results show that PANet effectively enriches existing hazy image benchmarks, significantly improving the performance of current dehazing models.
Primary Area: applications to computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 2323
Loading