Adaptive double PCA method of phase extraction in fringe projection profilometry

Abstract

To improve the reconstruction efficiency in the fringe projection profilometry, we present an adaptive double principal component analysis (DPCA) algorithm to remove the zero frequency and carrier frequency. The proposed DPCA method consists of the background removal PCA (BRPCA) and carrier removal PCA (CRPCA). The BRPCA algorithm is used to remove zero-frequency component and suppress the spectrum overlapping. The threshold is adaptively adjusted according to the change of the singular value and the matrix reconstruction dimension is adaptively obtained. The CRPCA algorithm is used to remove the carrier in the fundamental frequency component to obtain a phase map that contains only the height distribution of the tested object. Experiment results demonstrate that the proposed method can effectively remove zero frequency and carrier frequency and has higher reconstruction accuracy and anti-noise ability than existing state-of-the-art methods.