| 摘要 |
An MPEG decoder (100-180) can use different formats (linear tiled etc) for image data memory (150) for reference pictures when decoding subsequent pictures. Each format brings advantages and disadvantages. The disclosure illustrates that it is possible for different video sequences to have characteristics which match one image buffer format well, but not another, for example where there is a lot of motion or only a little motion between images. Advantageously, the decoder employs a parameterised image buffer format and analysis of the incoming stream and/or measured decoding performance (190) so as to dynamically choose (90, 160, 170) the best image buffer format for the current part of a stream. Cache control features can also be controlled (198) to suit the particular data received. |
