PnP-ReG: Learned Regularizing Gradient for Plug-and-Play Gradient Descent

The plug-and-play framework makes it possible to integrate advanced image denoising priors into optimization algorithms efficiently solve a variety of restoration tasks generally formulated as maximum posteriori (MAP) estimation problems. alternating direction method multipliers (ADMM) and the regularization by (RED) are two examples such methods that made breakthrough in restoration. However, former approach only applies proximal algorithms. And while explicit RED can be used various algorithms, including gradient descent, regularizer computed residual leads several approximations underlying prior MAP interpretation denoiser. We show is train network directly modeling jointly training corresponding use this gradient-based obtain better results compared other generic approaches. also pretrained for unrolled descent. Lastly, we resulting denoiser allows convergence ADMM.