摘要 |
A reconnaissance technique for the position of an object (8, 11) on successive images (7, 10) is obtained by pairing optimally points (A, B) and images by a single movement (.DELTA.m) compatible with the movement of the camera that has taken the two images. The position of the object may thus be calculated. The points that can be paired in practice belong to the object even if they have been obtained automatically, the background of the image often having a lower number of remarkable points and that it is impossible to pair with this movement (.DELTA.m).
|