| 摘要 |
<p>Motion information to be associated with pixel blocks of a current inter-version reference image (IVRI) (e.g. Inter Layer Reference images) to be inserted in a reference image buffer is derived. The buffer is used for temporal prediction of a current image belonging to a second version of a multi-version video (e.g. at least three scalable or multi-view versions). Pixel blocks of the current IVRI are derived from an first version image of the same video (e.g. which corresponds temporally to the current image). For a current pixel block of the current IVRI, a reference pixel block of the first version image which spatially corresponds to the current pixel block is determined, which is associated to first motion information. Properties of that first motion information are then obtained. Different motion information (8.10) (e.g. no motion, followed by setting the mode to INTRA, or using a different pixel block) is associated to the current pixel block if it is determined from the properties that the first motion information points to an IVRI held in the buffer which was used for temporal prediction of first version images. Artefacts introduced by the re-use of the same motion information twice are therefore avoided.</p> |
