| 摘要 |
<p>A method of determining the coordinates of feature points on the surface of an object from as few as one digital image generated by as few as a single 2-D sensor such as a camera is disclosed. At least three non-collinear points are detected among the images to generate at least three non-parallel 3-D lines. The pose of the surface is determined from these 3-D lines and a set of corresponding points on an ideal object. The points on the ideal object are transformed close to the 3-D lines in a least squares process. The other feature points on the surface can then be calculated by the intersections of the 3-D lines, generated from the detected feature points, with the surface at the pose determined above. The coordinates determined for these feature points are used in CRT measurements for picture geometry fine-tuning.</p> |
