|
Image Processing for the Analysis of Wave Interference Patterns in Physics: Review of Computational Methodologies
|
Mahdi Mirzaei   , Alireza Mohammadi |
|
|
|
Abstract: (1 Views) |
constitutes a fundamental pillar of precision measurement in both physics and engineering. These fringe patterns contain essential physical information such as displacement, surface topology, and refractive index. However, accurate phase extraction from noisy digitized intensity data relies heavily on computational power and advanced computer engineering techniques.
Classical interferometric approaches, particularly Phase-Shifting Interferometry (PSI) and Fourier Transform Profilometry (FTP), have long served as primary tools for phase retrieval, yet they face an intrinsic trade-off between accuracy and speed: PSI offers pixel-level precision and high stability but is slow and motion-sensitive, whereas FTP enables rapid measurement at the cost of reduced spatial resolution.
This review examines recent research on digital phase-retrieval algorithms and phase-unwrapping (PhU) methods from the perspectives of accuracy, robustness, computational cost, and generalizability. Findings indicate that deep learning (DL) has resolved the traditional accuracy–speed dilemma by enabling high-quality single-frame phase retrieval while providing remarkable resilience to noise [7, 4, 3, 2].
The core contribution of this paper is the introduction of a physics-informed deep learning framework (PI-FPA), which explicitly integrates prior physical knowledge into the network architecture and employs modules such as LeFTP to overcome the poor generalization of purely data-driven models .This framework significantly reduces the demand for large training datasets while improving accuracy, stability, and transferability, thereby establishing a new standard for fast and precise computational sensing in wave physics.
|
Article number: 4 |
| Keywords: Image Processing, Digital Interferometry, Phase Retrieval, Deep Learning, Physics-Informed Deep Learning, Wave Interference Patterns |
|
|
Full-Text [PDF 510 kb]
(3 Downloads)
|
Type of Study: Research |
Subject:
Special
Received: 2025/12/16 | Accepted: 2026/10/3 | Published: 2026/10/3
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
