In the phase measuring profilometry (PMP) three-dimensional measurement based on Stoilov algorithm, the four abnormal phenomena lead to the incorrect phase calculation or incorrect phase unwrapping, so the measured object can′t be reconstructed well. The proposed method based on the statistical approach can restrain the abnormal phenomena availably and make up Stoilov algorithm defect. Meanwhile, it is found that Soilov algorithm is sensitive to the period and contrast of sinusoidal gratings. The pixel of phase imaging will be mutant and the measuring precision is to be influenced when the light intensity varies severely. Therefore, a method of grating parameter optimization based on the statistical approach Soilov algorithm is proposed. The measuring precision can be improved by optimizing the period and contrast of sinusoidal gratings under the experimental simulation. Experimental result shows its feasibility and validity.