Bernoulli-guided multi-scale EKF method for motion error compensation in phase-shifting profilometry.

Area of Science:

• Optics and Photonics
• Computer Vision
• Metrology

Background:

• Motion artifacts are a significant challenge in optical metrology, particularly in 3D shape reconstruction.
• Existing phase shift estimation methods struggle with accuracy in dynamic scenes and at object edges.

Purpose of the Study:

• To develop a novel multi-scale Extended Kalman Filter (EKF)-based phase shift estimation technique.
• To compensate for motion-induced phase errors in real-time 3D shape reconstruction.
• To improve estimation accuracy, especially at the edges of measured objects.

Main Methods:

• Construction of a phase state space model using fringe information.
• Development of an EKF phase shift estimator utilizing the Bernoulli distribution.
• Implementation of a multi-scale sliding window for full-field fringe pattern traversal.
• Evaluation of actual phase shifts to minimize motion-induced errors.

Main Results:

• Enhanced estimation accuracy at object edges due to Bernoulli distribution guidance.
• Effective reduction of motion-induced phase errors.
• Demonstrated reliability in compensating for rotational and translational motion.
• Accurate 3D shape reconstruction achieved in dynamic scenes.

Conclusions:

• The proposed multi-scale EKF-based method offers a robust solution for motion compensation in phase-based 3D reconstruction.
• The technique provides accurate and reliable 3D shape measurements even under dynamic conditions.
• This advancement has significant implications for applications requiring precise 3D data acquisition in moving scenarios.