Single-Shot Multi-Frequency 3D Shape Measurement for Discontinuous Surface Object Based on Deep Learning.

Area of Science:

• Optics and Photonics
• Computer Vision
• Metrology

Background:

• Accurate absolute phase retrieval in fringe projection profilometry (FPP) is difficult with single-shot patterns.
• Existing methods struggle with stable and rapid measurements for isolated objects.

Purpose of the Study:

• To develop a single-shot method for fast and accurate absolute phase retrieval in FPP.
• To enable precise 3D shape measurement using only one fringe image.

Main Methods:

• A deep learning framework (DLFT) predicts unit-frequency and high-frequency fringe images from a low-frequency input.
• A second deep learning framework (DLPR) calculates the absolute phase using three predicted fringe images.
• The proposed method is termed single-shot multi-frequency absolute phase retrieval deep learning (SAPR-DL).

Main Results:

• The SAPR-DL method successfully retrieves absolute phase from a single fringe image.
• Experimental results demonstrate accurate 3D shape measurement of complex objects.
• The method achieves stable and rapid measurements.

Conclusions:

• The SAPR-DL method offers a significant advancement for single-shot FPP.
• This technique has broad potential for scientific and engineering applications requiring 3D shape analysis.