The ideal surface has strong polarization properties, but the actual surface is usually rough, resulting in significant shadowing and severe diffuse reflection. In order to accurately characterize the polarization properties of rough surfaces, based on the micro-facet model, an improved six-parameter polarization bidirectional reflection distribution function (pBRDF) model is established considering the shadowing effect and diffuse reflection. Based on this, the expression of the optical reflection polarization degree of rough surface is derived. The degree of polarization of aluminum and black lacquer materials is simulated to invert the complex refractive index. The results show that the proposed model simulation value is more consistent with the experimental measurement value, and the complex refractive index inversion accuracy is higher. The pBRDF measurement is carried out on the 45# steel material, and the distribution of polarization degrees under different incident angles reflection angles and roughness is analyzed. The experimental results show that the simulation values of the six-parameter model can be well matched with the experimental measurements, and the accuracy of the model is increased by modeling the diffuse reflection part. This work can be used as a theoretical basis and method for the detection and analysis of rough surfaces.