Local-polynomial-approximation-based phase unwrapping using state space analysis

Abstract

A new method of phase unwrapping is proposed for the absolute phase estimation based on the first-order polynomial phase approximation within a symmetric window generated around each pixel. The accurate estimation of polynomial coefficients is performed by formulating them as the elements of a state vector in a state space model. The extended Kalman filter offers a robust approach for the state estimation with the capability of handling high noise power. The polynomial coefficients estimates obtained at a given pixel are used as the initial conditions of the Kalman filter for its neighboring pixel, which results in the unwrapped phase estimation. The simulation and experimental results validate the performance of the proposed phase unwrapping method along with its ability of handling the masked phase fringe patterns with the help of a predefined binary mask and a pixel selection strategy.