Deep-learning-assisted composite polarized fringe projection profilometry for three-dimensional measurement of...

Area of Science:

• Optical Measurement
• Computer Vision
• Metrology

Background:

• Highlights on high-dynamic-range objects complicate accurate 3D surface measurement.
• Existing polarization coding methods reduce highlights but introduce nonlinear errors and lower signal-to-noise ratio.

Purpose of the Study:

• To develop an efficient and accurate 3D measurement method for high-dynamic-range objects with challenging surface properties.
• To overcome limitations of traditional polarization coding techniques in 3D reconstruction.

Main Methods:

• A novel deep-learning-assisted composite polarization fringe projection profilometry method was proposed.
• This approach integrates deep learning with physical-model-based polarization coding.
• The method aims to eliminate highlights, reduce nonlinear errors, and improve signal-to-noise ratio.

Main Results:

• Experimental results demonstrate significant improvements in measurement efficiency and accuracy.
• The proposed method effectively mitigates nonlinear errors and signal-to-noise ratio reduction.
• Validation across diverse scenes confirmed superior performance compared to traditional methods.

Conclusions:

• The deep-learning-assisted composite polarization method offers a robust solution for 3D measurement of high-dynamic-range objects.
• This technique enhances precision and speed in optical metrology for challenging surfaces.
• The integration of AI with physical models advances 3D reconstruction capabilities.