| 摘要 |
<p>The invention concerns a method whereby three-dimensional objects (1, 5, 6) of an environment which can be very complex can be observed by locating their simple contour forms on images, and then the geometrical characteristics, positions and orientations of said objects are measured. The contours are defined by simple parameters of length, inclination, position and the like. For each new image, the previously acquired knowledge of the environment is used to fine-tune the estimates while introducing into the model the new contours of the objects which have appeared.</p> |
