摘要 |
<p>The method involves initialising (41) the parameters which describe the movement of each in a number of regions (RI). A different movement hypotheses is selected after prediction errors are calculated, with the hypothesis having the least prediction errors chosen. This stage is followed by an intermediate processing stage (42) which provides filtering. A movement parameter refinement stage (43) has its output looped back to the input of the intermediate processing stage. An iterative function is performed until halted by predetermined criteria. This provides the final movement information (Mi(t)) corresponding to the region.</p> |