A Sequential Averaging Plug-and-Play Method for Image Restoration Via Fixed-Point Projection

Abstract

Plug-and-Play (PnP) methods are increasingly popular for solving image restoration problems. RED-PRO (regularization by denoising via fixed-point projection) is an important framework that bridges RED and PnP. Balancing the gradient method for data fidelity and the denoiser in RED-PRO with suitable stepsizes is a critical problem. To address this problem, we propose a Sequential Averaging Method via fixed-point projection called SAM-PRO, which is a special PnP method that achieves a reasonable tradeoff between the gradient method and the denoiser. Our method not only analyzes the equivalence and convergence rate of PnP methods but also achieves competitive or even superior performance compared to state-of-the-art PnP methods. The source code is available at https://github.com/zsc15/SAM-PRO.