prNet: Data-Driven Phase Retrieval via Stochastic Refinement

2025-07-13

AI Summary

To address the challenge of jointly optimizing distortion and perceptual quality in phase retrieval, this paper proposes the first model-driven stochastic sampling framework that integrates Langevin dynamics with deep denoising. Methodologically, it explicitly formulates the perceptual-distortion trade-off as a posterior distribution sampling problem, enabling synergistic optimization of data-driven priors and physical constraints via learnable noise scheduling, model-guided updates, and parallel Langevin inference. The framework supports warm-starting classical solvers while preserving both reconstruction fidelity and visual naturalness. Its key innovation lies in introducing perceptually oriented stochastic sampling to phase retrieval—marking a departure from conventional pixel-level error minimization paradigms. Evaluated on multiple benchmark tasks, the method achieves state-of-the-art performance, significantly improving structural fidelity and perceptual quality of reconstructed images.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

We propose a novel framework for phase retrieval that leverages Langevin dynamics to enable efficient posterior sampling, yielding reconstructions that explicitly balance distortion and perceptual quality. Unlike conventional approaches that prioritize pixel-wise accuracy, our method navigates the perception-distortion tradeoff through a principled combination of stochastic sampling, learned denoising, and model-based updates. The framework comprises three variants of increasing complexity, integrating theoretically grounded Langevin inference, adaptive noise schedule learning, parallel reconstruction sampling, and warm-start initialization from classical solvers. Extensive experiments demonstrate that our method achieves state-of-the-art performance across multiple benchmarks, both in terms of fidelity and perceptual quality.

Problem

• Develops phase retrieval balancing distortion and perception
• Uses stochastic sampling and denoising for reconstruction
• Achieves state-of-the-art fidelity and perceptual quality

Innovation

• Leverages Langevin dynamics for posterior sampling
• Combines stochastic sampling and learned denoising
• Integrates adaptive noise schedule learning