| 摘要 |
<p>The invention concerns a motion compensated prediction method for encoding digital data of a video image, the encoding comprising sectioning the image into blocks of images, the encoding of a residual block based on computation of the difference (2) between a current block and a predicted block, the predicted block being obtained after motion compensation (9) of one or several reconstructed blocks. The invention is characterised in that the current block and the predicted block are blocks of coefficients obtained after discrete cosine transform (1) of the image blocks and the computation (8) of the predicted coefficients of the predicted block of coefficients is carried out directly on the basis of the reconstructed coefficients of one or several blocks of reconstructed coefficients, this reconstruction being carried out from the addition (6) of one residual block to the corresponding predicted block, by weighting the reconstructed coefficients on the basis of the computed predicted coefficient.</p> |
