Blind Image Restoration

Blind image restoration aims to recover high-quality images from degraded versions where the type and parameters of the degradation are unknown. Current research heavily utilizes deep learning, focusing on diffusion models, transformer-based architectures, and plug-and-play methods incorporating learned denoisers, often incorporating techniques like prompt learning to enhance generalization across various degradation types. This field is crucial for improving the quality of images from various sources, including those captured by under-display cameras or affected by compression artifacts, impacting applications ranging from mobile photography to medical imaging.