| 摘要 |
A method of encoding/decoding an image, partitioned by boundaries into plural image portions (tiles) - each portion comprising one or more coding blocks, with one or more coding blocks of a second image portion being dependent, for encoding by a prediction process, on at least one coding block of a first image portion adjacent to the second image portion - comprises storing coding data of selected coding blocks of the first image portion for encoding or decoding one or more coding blocks of the second image portion. The coding blocks are selected, according to a subsampling pattern (e.g. adaptable pattern, based on coding block indexes or whether partition boundaries are horizontal or vertical), from the coding blocks of the first image portion bordering the (e.g. left side, vertical) partition boundary between the first image portion and the second image portion. Coding data may comprise motion data and/or intra-prediction data (e.g. planar or DC modes). Encoding data for unselected blocks may be based on that of selected blocks, e.g. determined to be equal to data of a selected block neighbouring the unselected block. |
