PG-DPIR: An efficient plug-and-play method for high-count Poisson-Gaussian inverse problems
Maud Biquard, Marie Chabert, Florence Genin, Christophe Latry, Thomas Oberlin
TL;DR
This work tackles high-count Poisson-Gaussian image restoration by introducing PG-DPIR, a plug-and-play method derived from DPIR that accommodates Poisson-Gaussian noise through a Gaussian approximation and a novel, efficient initialization for the proximal step. By initializing the inner gradient descent at a closed-form approximate proximal operator and splitting iterations into two phases, PG-DPIR achieves state-of-the-art restoration quality while dramatically reducing computation time. The approach demonstrates superior performance on realistic satellite image simulations (Pléiades-like), offering a practical, sensor-agnostic tool for on-ground processing pipelines. The results highlight PG-DPIR's potential to replace heavier optimization pipelines with a fast, robust PnP framework for Poisson-Gaussian inverse problems.
Abstract
Poisson-Gaussian noise describes the noise of various imaging systems thus the need of efficient algorithms for Poisson-Gaussian image restoration. Deep learning methods offer state-of-the-art performance but often require sensor-specific training when used in a supervised setting. A promising alternative is given by plug-and-play (PnP) methods, which consist in learning only a regularization through a denoiser, allowing to restore images from several sources with the same network. This paper introduces PG-DPIR, an efficient PnP method for high-count Poisson-Gaussian inverse problems, adapted from DPIR. While DPIR is designed for white Gaussian noise, a naive adaptation to Poisson-Gaussian noise leads to prohibitively slow algorithms due to the absence of a closed-form proximal operator. To address this, we adapt DPIR for the specificities of Poisson-Gaussian noise and propose in particular an efficient initialization of the gradient descent required for the proximal step that accelerates convergence by several orders of magnitude. Experiments are conducted on satellite image restoration and super-resolution problems. High-resolution realistic Pleiades images are simulated for the experiments, which demonstrate that PG-DPIR achieves state-of-the-art performance with improved efficiency, which seems promising for on-ground satellite processing chains.