Neural network-based phase extraction method for digital moiré fringes in single-grating displacement measurement...

Area of Science:

• Optics and Photonics
• Machine Learning Applications
• Metrology

Background:

• Digital moiré fringe patterns are crucial for displacement measurement.
• Traditional phase extraction methods can suffer from spectral leakage and require complex algorithms.
• Single-grating systems offer a cost-effective approach to displacement measurement.

Purpose of the Study:

• To develop a novel, deep learning-based phase extraction method for digital moiré fringe patterns.
• To improve accuracy and robustness in nanoscale displacement measurements.
• To eliminate errors caused by spectral leakage in single-grating systems.

Main Methods:

• Artificial generation of digital moiré fringes by superimposing a 2D cosine image onto grating images.
• Ensuring integer periods of moiré fringes in sampled data to prevent spectral leakage.
• Utilizing a trained neural network for phase detection, bypassing complex signal decomposition.

Main Results:

• The proposed deep learning method achieves superior phase extraction accuracy compared to variational mode decomposition.
• The method demonstrates enhanced robustness in displacement measurements.
• Successful implementation in a single-grating configuration for nanoscale measurements.

Conclusions:

• The novel deep learning approach provides a low-cost, high-precision solution for nanoscale displacement measurement.
• Artificial moiré fringe generation effectively eliminates phase extraction errors.
• The method simplifies the measurement process by avoiding complex algorithms and parameter tuning.