| 摘要 |
A process burden for code processing is reduced by the present system and method while possibly avoiding the lowering of image quality. When compensating for motion prediction by using a multi-reference frame while sequentially changing the pixel-based sizes of motion compensating blocks, a size-reduced block is generated depending upon a motion compensating block having the greatest pixel-based size to be taken as the uppermost layer among the pixel-based sizes. Thereafter, motion vector search ranges are determined respectively within the reference frame images, on the basis of a plurality of size-reduced reference images reduced in size respectively corresponding to the size-reduction ratios of the size-reduced blocks and the size-reduced blocks. Using the determined search range, an optimal motion vector is detected while sequentially changing the pixel-based sizes of the motion compensating blocks. This can detect a motion vector by only a limited search range within the reference frame image.
|
