FIELD: physics.
SUBSTANCE: device for decoding a moving image, adapted to decoding of the decoding block consisting of, at least, one prediction block, contains a decoding block adapted to decode a candidate determinig index from a bitstream, in which the candidate determinig index with information on the merge movement used in the prediction block, that is a decoding object, is encoded as the candidate determinig index; a candidate generating block with information on the merge movement, adapted to retrieve the set of candidates with information on the merge movement shared for all prediction blocks in the decoding block, implemented in case, when information indicating the extraction or non-extraction of the candidate with information on the merge movement shared for all prediction blocks in the decoding block, is information indicating the extraction of the candidate with information on the merge movement shared for all prediction blocks in the decoding block; and a block selecting information on the merge movement, adapted to select one candidate with information on the merge movement from the set of candidates with information on the merge movement and using the selected candidate with information on the merge movement in the prediction block, that is a decoding object, wherein the candidate generating block with information on the merge movement removes a lot of candidates with information on the merge movement on the basis of the size and position of the prediction block , that is a decoding object, in case, when information pointing to the extraction or non-extraction of the candidate with information on the merge movement shared for all prediction blocks in the decoding block, is information indicating non-extraction of the candidate with information on the merge movement shared for all prediction blocks in the decoding block.
EFFECT: creating an effective technology for decoding a moving image, that allows to provide high efficiency of parallel motion-compensated prediction data processing while simultaneously performing motion-compensated prediction on a plurality of contiguous prediction blocks.
3 cl, 36 dwg
