CurvPnP: Plug-and-play blind image restoration with deep curvature denoiser

Abstract

Due to the development of deep learning-based denoisers, the plug-and-play strategy has achieved great success in image restoration problems. However, existing plug-and-play image restoration methods are designed for non-blind Gaussian denoising such as Zhang et al. (2022), the performance of which visibly deteriorates for unknown noise. To push the limits of plug-and-play image restoration, we propose a novel image restoration framework with a blind Gaussian prior, which can deal with more complicated image restoration problems in the real world. More specifically, we build up a curvature regularization image restoration model by regarding the noise level as a variable, where the regularization term is realized by a two-stage blind Gaussian denoiser consisting of a noise estimation subnetwork and a denoising subnetwork. We also introduce curvature regularization into the encoder–decoder architecture and the supervised attention module to achieve a highly flexible and effective network. Numerous experimental results are provided to demonstrate the advantages of our deep curvature denoiser and the resulting plug-and-play blind image restoration method over the state-of-the-art denoising methods. Our model is shown to be able to recover fine image details and tiny structures even when the noise level is unknown for different image restoration tasks. The source codes are available at https://github.com/Duanlab123/CurvPnP.