Background-extracted extended Kalman filter-based phase shift estimation algorithm for phase shifting profilometry system

Abstract

The accuracy of phase shift in the phase-measuring profilometry (PMP) directly affect the result of the three-dimensional reconstruction. A robust phase shifting estimation method based on the extended Kalman filter (EKF) is proposed to address the imprecise phase shift induced by the mechanical projector and dynamic object. Unlike the traditional fringe projection profilometry (FPP) in which the zero frequency is suppressed and the fundamental frequency is extracted, the zero-frequency component is extracted by using two-dimensional variational mode decomposition and bidimensional empirical mode decomposition (2D-VMD-BEMD) algorithm, and is used to establish the predicted observations of the measurement process in the EKF. The coefficient of the quadratic phase in a selected window is employed to estimate the actual phase step by using EKF and subsequently the accurate reconstructed phase map is obtained. This paper develops a novel 2D-VMD-BEMD EKF estimator to evaluate the actual phase shift and enhance the accuracy of the phase reconstruction, and simultaneously provides a reliable and effective method for the extraction of background under unfavorable conditions. Simulation and experimental results demonstrate that the proposed region-wise estimation method effectively removes the effect of the unexpected phase shift, and can be widely available for N-step phase shifting algorithm in phase shifting profilometry.