3D shape measurement method for multi-reflective scenes based on accurate pixel-by-pixel phase-error adjustment strategy

•A reflection model and an error transformation model are built to analyze the disturbances induced by MRVs in the captured fringe patterns.•The exposure time range of each pixel can be calculated pixel-by-pixel by setting a proper phase error threshold.•An efficient way to globally minimize the number of exposures based on the greedy algorithm is introduced.•Highly accurate point cloud integration (APCI) is adjusted to evaluate the reconstruction accuracy pixel by pixel, which can indicate the overall integrity of the highly-accurate point cloud. Fringe projection profilometry (FPP) has been widely applied to three-dimensional (3D) shape measurement with the advantages of non-contact and high accuracy. However, measuring the scenes of multi-reflective variations (MRVs) is still a tricky task for conventional FPP because the dark-field and saturation could lead to phase errors and reconstruction errors. In this paper, a novel high dynamic range (HDR) technique based on accurate pixel-by-pixel phase-error adjustment strategy is proposed, which is mainly focus on both accuracy and quality of the reconstruction. A reflection model and an error transformation model are built to analyze the disturbances induced by MRVs in the captured fringe patterns. By setting a proper phase error threshold, the exposure time range of each pixel can be calculated. Finally, a greedy algorithm is introduced to globally minimize the number of exposures. Experiments results have verified that our proposed method can improve both the measurement accuracy and accurate point cloud integrality (APCI) compared with conventional FPP and several HDR methods, making it suitable to be further applied in the HDR 3D shape measurement.