3D vision can realize accurate statistics of passenger flow with higher accuracy and lower cost than 2D vision under unguaranteed environmental lighting conditions and crowded scene, and can realize more abnormal behavior analysis based on 3D data. It is suitable for difficult scenes that the traditional 2D visual scheme is not competent for, such as vehicles, stations, airports, exhibition halls, scenic spots, schools, stadiums and other crowded places, or security-sensitive places such as banks.