Generative Image Denoising

Generative image denoising aims to reconstruct clean images from noisy versions using generative models, primarily focusing on improving the quality and efficiency of this reconstruction. Current research emphasizes the use of diffusion models, often integrated within plug-and-play frameworks or enhanced with residual connections, to achieve state-of-the-art results across various image restoration tasks. These advancements are significant because they offer improved image quality and scalability for applications such as snapshot compressive imaging and other inverse problems, pushing the boundaries of image processing and computer vision.