Aiming at the problems of fringe modulation degradation and phase unwrapping relying on extra coding and being easily affected by fringe quality in high dynamic range (HDR) scenes in 3D measurement of HDR objects, this paper proposes an HDR object 3D measurement method (G-PPSF) fusing polarization phase-shifting modulation and geometric prior constraints. Based on the Stokes-Mueller polarization optics theory, this method constructs a polarization phase-shifting projection model under extremely small incident angle, and designs a (2N + 2)-frame compact projection sequence. It can suppress fringe distortion caused by surface reflectivity difference with a single exposure, and achieve high-quality wrapped phase acquisition in HDR scenes. Geometric prior constraints are introduced to construct a virtual reference plane, which is independent of fringe intensity information and inherently robust to HDR scenes, enabling absolute phase unwrapping without projecting extra auxiliary fringes. Experimental results show that compared with traditional methods, G-PPSF can maintain a point cloud data integrity rate of more than 97% even under high noise filtering thresholds. In the plane fitting test, its root mean square error (RMSE) is stable at 0.0773 mm, and the effective data ratio is as high as 99.1%. This method does not require multi-exposure fusion or complex iterative calculation, and realizes efficient, robust and complete 3D reconstruction in HDR scenes.
Wang et al. (Thu,) studied this question.