Reproduce “HNDiff: Haze-Noise Diffusion for Image Dehazing” 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
Jin-Ting He, Fu-Jen Tsai, Yan-Tsung Peng, Min-Hung Chen, Chia-Wen Lin, Yen-Yu Lin
HNDiff restores hazy images by aligning diffusion with atmospheric scattering rather than treating degradation as random noise alone. Its forward process adds haze and Gaussian noise, while a haze-density-aware schedule varies noise across the scene and learned estimators reverse both. On NH-HAZE, FocalNet reaches 20.89 dB PSNR, up from 20.36, across five test pairs; it is designed for atmospheric scattering.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.