Fringe projection profilometry (FPP) has been widely used in various fields because of its high precision and efficiency. However, fringe images containing overexposed areas are obtained by traditional FPP for high-reflective object surfaces, which leads to errors in absolute phase and three-dimensional (3D) reconstruction results. A robust 3D reconstruction technique based on a polarization state-based ternary polarization coding strategy is proposed to address these errors. Instead of the light intensity coding patterns, the polarization coding patterns are used in this method. Next, the S₁ image is calculated using the polarization image captured by the camera. The Stokes parameter is utilized to segment the fringe codewords without using a specially designed threshold. The proposed method enhanced the stability of the fringes, improved the quality of the fringes, and reduced the number of fringe patterns. Experimental results verified that the proposed method has higher robustness than the conventional method in the measurement of objects with high-reflective surfaces. The results reveal that the proposed method can suppress the effect of highlight areas in the reconstruction of high-reflective surfaces to obtain accurate absolute phases and reconstruct the 3D point cloud data reliably.