Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep learning phase recovery: data-driven, physics-driven, or combining both?

Kaiqiang Wang, Edmund Y. Lam

TL;DR

This work comprehensively compare these two deep learning phase recovery strategies in terms of time consumption, accuracy, generalization ability, ill-posedness adaptability, and prior capacity and proposes a co-driven strategy of combining datasets and physics for the balance of high- and low-frequency information.

Abstract

Phase recovery, calculating the phase of a light wave from its intensity measurements, is essential for various applications, such as coherent diffraction imaging, adaptive optics, and biomedical imaging. It enables the reconstruction of an object's refractive index distribution or topography as well as the correction of imaging system aberrations. In recent years, deep learning has been proven to be highly effective in addressing phase recovery problems. Two most direct deep learning phase recovery strategies are data-driven (DD) with supervised learning mode and physics-driven (PD) with self-supervised learning mode. DD and PD achieve the same goal in different ways and lack the necessary study to reveal similarities and differences. Therefore, in this paper, we comprehensively compare these two deep learning phase recovery strategies in terms of time consumption, accuracy, generalization ability, ill-posedness adaptability, and prior capacity. What's more, we propose a co-driven (CD) strategy of combining datasets and physics for the balance of high- and low-frequency information. The codes for DD, PD, and CD are publicly available at https://github.com/kqwang/DLPR.

Deep learning phase recovery: data-driven, physics-driven, or combining both?

TL;DR

This work comprehensively compare these two deep learning phase recovery strategies in terms of time consumption, accuracy, generalization ability, ill-posedness adaptability, and prior capacity and proposes a co-driven strategy of combining datasets and physics for the balance of high- and low-frequency information.

Abstract

Phase recovery, calculating the phase of a light wave from its intensity measurements, is essential for various applications, such as coherent diffraction imaging, adaptive optics, and biomedical imaging. It enables the reconstruction of an object's refractive index distribution or topography as well as the correction of imaging system aberrations. In recent years, deep learning has been proven to be highly effective in addressing phase recovery problems. Two most direct deep learning phase recovery strategies are data-driven (DD) with supervised learning mode and physics-driven (PD) with self-supervised learning mode. DD and PD achieve the same goal in different ways and lack the necessary study to reveal similarities and differences. Therefore, in this paper, we comprehensively compare these two deep learning phase recovery strategies in terms of time consumption, accuracy, generalization ability, ill-posedness adaptability, and prior capacity. What's more, we propose a co-driven (CD) strategy of combining datasets and physics for the balance of high- and low-frequency information. The codes for DD, PD, and CD are publicly available at https://github.com/kqwang/DLPR.
Paper Structure (11 sections, 12 equations, 13 figures, 2 tables)

This paper contains 11 sections, 12 equations, 13 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Principles and Methods
  3. Results and discussion
  4. Comparison of time consumption and accuracy
  5. Comparison of generalization abilit
  6. Comparison of ill-posedness adaptability
  7. Comparison of prior capacity
  8. Comparison of experimental data
  9. Conclusion
  10. Dataset generation
  11. Network implementation

Figures (13)

  • Figure 1: Phase recovery network training with data-driven and physics-driven strategies.
  • Figure 2: Description of dataset-driven deep learning phase recovery methods.
  • Figure 3: Description of physics-driven deep learning phase recovery methods. (a) Network inference for the uPD. (b) Network training and inference for the tPD. (c) Network training and inference for the tPDr.
  • Figure 4: Inference results of DD, uPD, tPD, and tPDr.
  • Figure 5: Results of DD, tPD, and co-driven. The blue box represents low-frequency information and the green box represents high-frequency information
  • ...and 8 more figures