Toward Generalized and Realistic Unpaired Image Dehazing via Region-Aware Physical Constraints
- 13 November 2024
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 35 (3) , 2753-2767
Abstract
Supervised dehazing models, trained on synthetic hazy-clean image pairs, often face a notable decline in performance when applied to real-world scenes. Consequently, CycleGAN-based unpaired dehazing methods are proposed to improve the model’s generalization. One successful approach among these methods involves decomposing the physical properties of the atmospheric scattering model (ASM). However, estimating physical properties individually from input images is difficult without supervised labels, which ignores the semantic consistency between different physical regions. We claim semantic region information can offer additional geometric spatial constraints for estimating physical properties, as natural images can be divided into regions with similar scene depths. Motivated by this, we propose a novel generalized and realistic unpaired image dehazing framework via region-aware physical constraints (RPC-Dehaze). Our approach utilizes fine-grained semantic region maps from the Segment Anything Model (SAM) in a specially designed region prompt enhancement module. This enables the dehazing and hazing cyclic networks to learn region-aware physical constraints, leading to accurate estimation of haze imaging physical properties. In contrast to existing unpaired methods that treat dehazing and hazing networks equally, we incorporate Retinex theory into the hazing network, allowing it to learn diverse illumination effects in different regions. We adaptively refine the Retinex-based illumination component, resulting in more realistic hazy images. To further facilitate unsupervised learning in our framework, we propose a physical consensual contrastive regularization to ensure compact representation constraints in the latent feature space. Extensive experiments on synthetic and real image datasets show our method surpasses state-of-the-art unpaired dehazing methods in both effectiveness and generalization capability.Keywords
Funding Information
- National Natural Science Foundation of China (U23B2011)
- National Natural Science Foundation of China (62102069)
- National Natural Science Foundation of China (U20B2063)
- National Natural Science Foundation of China (62220106008)
- Sichuan Science and Technology Program (2022YFG0032)
- China Academy of Space Technology Innovation Program
This publication has 59 references indexed in Scilit:
- Robust Unpaired Image Dehazing via Density and Depth DecompositionInternational Journal of Computer Vision, 2023
- SG-Net: Semantic Guided Network for Image Dehazing2023
- Unpaired Deep Image Dehazing Using Contrastive Disentanglement Learning2022
- You Only Look Yourself: Unsupervised and Untrained Single Image Dehazing Neural NetworkInternational Journal of Computer Vision, 2021
- Context-Enhanced Representation Learning for Single Image DerainingInternational Journal of Computer Vision, 2021
- Self-supervised Monocular Depth Estimation: Solving the Dynamic Object Problem by Semantic Guidance2020
- I-HAZE: A Dehazing Benchmark with Real Hazy and Haze-Free Indoor Images2018
- Single Image Dehazing via Multi-scale Convolutional Neural Networks2016
- Single image dehazing via an improved atmospheric scattering modelThe Visual Computer, 2016
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015