Virtual temporal phase-shifting phase extraction using generative adversarial networks.

Area of Science:

• Optical metrology
• Artificial intelligence in imaging

Background:

• Dynamic optical measurement often requires multiple fringe patterns.
• Traditional phase-shifting methods can be complex and sensitive to environmental noise.

Purpose of the Study:

• To introduce a novel virtual temporal phase-shifting method (PSNet) for dynamic measurements.
• To leverage generative adversarial networks (GANs) for enhanced fringe pattern analysis.

Main Methods:

• PSNet generates virtual phase-shifting fringe patterns from a single captured fringe pattern.
• A standard n-step phase-shifting algorithm is applied to the virtual patterns.
• Phase unwrapping is performed using established algorithms.

Main Results:

• PSNet successfully generates noise-free virtual phase-shifting fringe patterns.
• The denoising step is eliminated due to the inherent noise reduction.
• Experimental validation using captured fringe patterns in an interferometer confirms the method's effectiveness.

Conclusions:

• The proposed PSNet method offers a practical and efficient solution for dynamic measurements.
• GANs can effectively synthesize virtual phase-shifting patterns, simplifying optical measurement processes.
• The method demonstrates high performance and robustness in real-world interferometric applications.