Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network

2024-06-12

Neural Information Processing Systems

AI Summary

To address the challenges of image dehazing under complex, real-world haze distributions and scarce paired training data, this paper proposes a physics-guided collaborative unfolding network coupled with a consistency-driven pseudo-label generation framework. The method introduces three key innovations: (1) the first iterative Mean-Teacher-based pseudo-label generator specifically designed for real-world image dehazing (RID); (2) a collaborative unfolding architecture explicitly incorporating the atmospheric scattering model; and (3) a global-local consistency mechanism that dynamically constructs a weighted pseudo-label pool to prioritize haze-free region recovery. It further integrates physics-constrained loss, multi-scale feature fusion, and weight-adaptive regularization. Extensive experiments demonstrate state-of-the-art performance across multiple real-world dehazing benchmarks, with significant improvements in detail reconstruction fidelity and color accuracy. The source code is publicly available.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Humans and AI: Human-in-the-loop Machine Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Data and user privacy-enhancing technologies for the Web
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Real-world Image Dehazing (RID) aims to alleviate haze-induced degradation in real-world settings. This task remains challenging due to the complexities in accurately modeling real haze distributions and the scarcity of paired real-world data. To address these challenges, we first introduce a cooperative unfolding network that jointly models atmospheric scattering and image scenes, effectively integrating physical knowledge into deep networks to restore haze-contaminated details. Additionally, we propose the first RID-oriented iterative mean-teacher framework, termed the Coherence-based Label Generator, to generate high-quality pseudo labels for network training. Specifically, we provide an optimal label pool to store the best pseudo-labels during network training, leveraging both global and local coherence to select high-quality candidates and assign weights to prioritize haze-free regions. We verify the effectiveness of our method, with experiments demonstrating that it achieves state-of-the-art performance on RID tasks. Code will be available at https://github.com/cnyvfang/CORUN-Colabator.

Problem

• Addressing haze-induced degradation in real-world images
• Modeling real haze distributions without paired data
• Generating high-quality pseudo labels for dehazing networks

Innovation

• Cooperative unfolding network integrates physical knowledge
• Coherence-based Label Generator produces pseudo labels
• Optimal label pool prioritizes haze-free regions