In three-dimensional (3D) scanning via phase measuring profilometry, if there exists saturation in captured patterned images, the accuracy of reconstructed 3D point clouds will be ruined. The phase error caused by saturated pixel values is studied, and a mathematical model for the error is derived. With such a model, a novel algorithm is proposed to minimize the phase error by employing unsaturated pixel values. The proposed method is proved by both simulation and experiment, i.e., analysis of the root mean square error of the corrected phase will show a 92.5% reduction in phase error in simulation, and an 82.8% reduction in phase error in the experiment.