When deep denoising meets iterative phase retrieval

TL;DR

This paper introduces a hybrid approach combining iterative phase retrieval algorithms with deep denoisers, significantly improving noise robustness and performance in Fourier intensity-based signal recovery tasks.

Contribution

It presents a novel regularization-by-denoising framework that integrates deep denoisers into iterative phase retrieval, enhancing robustness and outperforming existing methods.

Findings

Outperforms existing noise-robust phase retrieval algorithms

Combines advantages of iterative methods and deep denoisers

Enables hybrid imaging methods with learned constraints

Abstract

Recovering a signal from its Fourier intensity underlies many important applications, including lensless imaging and imaging through scattering media. Conventional algorithms for retrieving the phase suffer when noise is present but display global convergence when given clean data. Neural networks have been used to improve algorithm robustness, but efforts to date are sensitive to initial conditions and give inconsistent performance. Here, we combine iterative methods from phase retrieval with image statistics from deep denoisers, via regularization-by-denoising. The resulting methods inherit the advantages of each approach and outperform other noise-robust phase retrieval algorithms. Our work paves the way for hybrid imaging methods that integrate machine-learned constraints in conventional algorithms.