| 摘要 |
In this system, three-dimensional objects ( 1, 5, 6 ) in an environment that may be very complex are discerned by identifying their simple contour shapes on images, and the geometric characteristics of these objects and their positions and orientations are then measured. Contours are defined by simple length, inclination, position parameters, etc. The acquired knowledge of the environment is used in each new image to refine the estimates, while introducing new object contours that have appeared, into the model. |
