Movement estimation process and device

摘要

The matrix network (3) is used to calculate an error function between a current block of pixels of an image for compression and blocks of the same size of a past or future reference image corresponding to a study window centred around the current block. The network comprises lines of cells (S) connected in series by buffers (B) each column being associated through buffers with adding units (Add). The adding units are also connected in series through three buffers, one of which may be short circuited by a differentiating circuit (301). The network also includes a final differentiating unit (302).