In order to identify cement road surface with different roughness by laser scattering characteristic, a road surface roughness measurement system was designed. The height distribution of cement road surface was measured by micrometer, and the root mean square of height and relevant length were calculated. According to the above parameters, the random rough surface was generated by power spectral frequency domain filtering to simulate cement road surface, and the rough surface was dispersed into a large number of surface elements by tangential plane approximation. The local field was calculated by Fresnel formula, and the Monte Carlo method was used to obtain the statistical mean values of bidirectional and the back scattered light intensity on rough surface under different polarized light incident conditions. Based on the virtual instrument technology, the high precision automatic laser scattering measurement was carried out, and then the theoretical model was verified according to the experimental data. The results show that the scattered light intensity of bidirectional scattered cement road surface with small roughness has a peak value of 90lx near the scattering angle of 40° in the mirror direction, and decreases gradually on both sides of the mirror direction. The optical characteristics of cement road surface with large roughness are approximate to Lambert body, and the scattered light intensity has little change in each scattering angle direction. The back scattered light intensity peak of small roughness cement road surface is 103lx when the back scattering is perpendicular to the incidence, which decreases gradually with the increase of scattering angle. The large roughness cement road surface has Lambert scattering characteristics. The above conclusions could provide reference for road information perception of autonomous driving schemes.