| 摘要 |
The method involves receiving an image of a vehicle with a camera and determining the position and orientation of a perspective of the vehicle. The rendered two dimensional views (2D) are generated from three dimensional (3D) vehicle models stored in a database in positions and perspective orientations. The captured image of the vehicle is compared with the rendered 2D views of the stored 3D vehicle models for finding the 2D view with the closest match. The recorded vehicle is classified with the 3D vehicle model that is best matched with one of the 2D views. |
