Point-Wise Phase Estimation Method in Fringe Projection Profilometry under Non-Sinusoidal Distortion.

Area of Science:

• Metrology
• Optical Engineering
• Computer Vision

Background:

• Fringe projection profilometry is sensitive to non-sinusoidal fringe distortions.
• High-order harmonics in distorted fringes introduce significant errors in phase estimation.
• Existing methods struggle to accurately resolve phase under complex distortions.

Purpose of the Study:

• To propose a novel point-wise phase estimation method based on a neural network (PWPE-NN).
• To address and eliminate phase estimation errors caused by non-sinusoidal fringe distortions.
• To establish an implicit phase solution without complex measurement operations.

Main Methods:

• A simple neural network model is employed to construct the complex nonlinear mapping between gray values and phase.
• The method performs point-wise calculations for accurate phase value determination.
• It avoids the need for combining local image information, unlike traditional approaches.

Main Results:

• The proposed PWPE-NN method effectively eliminates phase errors stemming from non-sinusoidal phase shifting.
• Traditional methods exhibit periodic phase errors, which are absent in the PWPE-NN approach.
• Accurate phase calculation is achieved for each individual point.

Conclusions:

• The PWPE-NN method offers a robust solution for accurate phase estimation in fringe projection profilometry.
• It overcomes the limitations of traditional methods in handling non-sinusoidal distortions.
• This technique provides a more precise and reliable approach to 3D surface reconstruction.