Denoising diffusion post-processing for low-light image enhancement

Savvas Panagiotou, Anna S. Bosman

Abstract: Highlights•Low-light enhancement results in noisy images, treated by post-processing denoising.•Mapping from poor- to well-lit images can be captured by a conditional distribution.•Diffusion model is used to learn the low-light to well-lit conditional distribution.•Proposed diffusion model avoids the iterative process typical of diffusion models.•Our diffusion model achieves state-of-the-art results on post-processing denoising.