PG-DPIR: An efficient plug-and-play method for high-count Poisson-Gaussian inverse problems

2025-04-14

AI Summary

This work addresses image restoration under high-count Poisson–Gaussian mixed noise—typical in satellite remote sensing, where noise is signal-dependent and nonlinear. We propose an efficient plug-and-play (PnP) framework. Methodologically, we introduce the first DPIR-accelerated adaptation for the Poisson–Gaussian likelihood, incorporating a gradient-descent initialization strategy that accelerates convergence of the proximal step—lacking a closed-form solution—by several orders of magnitude. Our architecture employs a deep-unfolding PnP optimization scheme that tightly couples a learnable CNN denoiser with accurate noise modeling, augmented by adaptive forward–backward iterations and fast proximal solvers. Evaluated on Pleiades high-resolution satellite imagery, the method achieves state-of-the-art performance in both restoration and super-resolution, with significantly improved computational efficiency—demonstrating strong potential for onboard or ground-based real-time processing.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Mixed Discrete/Continuous Search
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Security and Privacy: Privacy-enhancing technologies

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.

Problem

• Develop efficient plug-and-play method for Poisson-Gaussian image restoration
• Address slow convergence in high-count Poisson-Gaussian inverse problems
• Achieve state-of-the-art performance in satellite image restoration

Innovation

• Plug-and-play method for Poisson-Gaussian noise
• Efficient initialization accelerates gradient descent
• State-of-the-art performance in image restoration