Polarization-based structured light 3D measurement for metallic surfaces under time-varying ambient illumination

In industrial measurement, the high reflectivity of metallic surfaces and the complex, time-varying ambient light interference (TVALI) are two primary factors that limit the accuracy of fringe projection profilometry (FPP). The former often causes local saturation and reduces fringe contrast, whereas the latter introduces a time-varying background component that compromises the accuracy of phase retrieval. To address these challenges, we propose a polarization-modulated illumination recovery model (P-MIRM) based on the physical mechanism of Fresnel reflection on metallic surfaces. By taking advantage of the strong polarization-preserving property of linearly polarized projected fringes on metallic surfaces, we adopt the response differences among the four channels of a polarization camera to achieve pixel-wise decoupling of time-varying ambient illumination from the modulated signals using a single image. Compared with conventional schemes that rely on mechanical rotation, the proposed method requires no additional operations, which substantially simplifies system calibration and improves measurement efficiency. Experimental results on the metal standard piece show the proposed method reduces the standard deviation of plane-fitting residuals and the error of adjacent step height differences by approximately 73% and 47%, respectively, demonstrating markedly improved robustness under unstable illumination.