Reproduce “4KDehazeFlow: Ultra-High-Definition Image Dehazing via Flow Matching” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xingchi Chen, Pu Wang, Xuerui Li, Chaopeng Li, Juxiang Zhou, Jianhou Gan, Dianjie Lu, Guijuan Zhang and 2 more
4KDehazeFlow introduces a method for ultra-high-definition image dehazing using a Flow Matching framework that balances perceptual quality and computational efficiency. It achieves state-of-the-art performance on UHD datasets like 4KID, I-HAZE, and O-HAZE, demonstrating notably lower computational overhead compared to existing approaches.
Key insights
- Flow Matching offers an efficient and stable generative modeling paradigm for image restoration, avoiding the complex, lengthy iterative sampling of diffusion models while providing continuous optimization.
- The combination of a CNN-based atmospheric scattering purifier and an adaptive Haze-LUT effectively addresses both haze removal and critical color fidelity preservation, preventing common color distortions in dehazed UHD images.
- Utilizing a robust ODE solver like RK4 is crucial for maintaining structural and perceptual quality during the continuous transformation, ensuring a smooth and artifact-free dehazing process.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.