Related Experiment Video
Updated: Dec 25, 2025
11:34
High-resolution, High-speed, Three-dimensional Video Imaging with Digital Fringe Projection Techniques
Published on: December 3, 2013
16.0K
Dynamic 3-D measurement based on fringe-to-fringe transformation using deep learning
Optics Express
|April 1, 2020
Summary
This study introduces a deep learning method for faster 3-D shape measurement using fringe projection profilometry (FPP). The technique reduces motion errors in dynamic objects by requiring fewer fringes for accurate phase retrieval.
Area of Science:
- Optics and Photonics
- Computer Vision
- Metrology
Background:
- Fringe projection profilometry (FPP) is crucial for dynamic 3-D shape measurement.
- Traditional phase retrieval in FPP often needs many fringes, causing errors with moving objects.
Purpose of the Study:
- To develop a novel, deep learning-based phase retrieval technique for FPP.
- To minimize motion-induced errors in dynamic 3-D measurements.
Main Methods:
- An end-to-end deep convolution neural network (CNN) was designed.
- The CNN transforms one or two fringe patterns into phase-retrieval-ready data.
- Input fringe requirements vary based on object depth (single fringe for restricted depth, two for unrestricted).
Main Results:
- The deep learning approach significantly reduces the number of fringes needed for phase retrieval.
- The method demonstrates applicability for dynamic 3-D measurements.
- Theoretical analysis, numerical simulations, and experimental validation confirm the technique's effectiveness.
Conclusions:
- The proposed deep learning technique offers an efficient solution for phase retrieval in dynamic FPP.
- This advancement enables more accurate and faster 3-D shape measurement of moving objects.
