Single Image Denoising

Single image denoising aims to remove noise from a single image without a clean reference, a crucial task in computer vision with applications ranging from medical imaging to photography. Recent research emphasizes developing robust methods for handling real-world noise, which is often non-Gaussian and spatially varying, moving beyond simpler additive white Gaussian noise models. This involves exploring diverse architectures, including non-local methods, deep learning approaches (like diffusion models and transformers), and self-supervised learning techniques that leverage the noisy image itself for training. These advancements improve image quality and enable more accurate downstream analyses in various fields.