| 摘要 |
The method for estimating the motion according to the invention calculates a priori for the current image a first motion field for blocks of a first level resulting from a partition of the image into blocks of so-called ''maximum'' size, from a symmetrical estimation algorithm, by using the preceding and following images in the sequence to calculate the displaced interaster differences, and a second motion field for blocks resulting from the partition of one block of the first level and so on up to a level of blocks of minimum size. The final motion field results from the selection of the best adapted motion vectors amongst those of different levels and is directly associated with a ''quadtree'' coding tree. The invention applies to image transmission channels, particularly in high resolution. |
