| 摘要 |
Apparatus for compressing a motion picture signal divided into blocks, in which circuits perform the following operations: subtracting the blocks of the motion picture signal from corresponding prediction blocks of a prediction picture to provide prediction error blocks. Orthogonally transforming the error blocks to provide transform coefficients. Quantizing the transform coefficients to provide quantized transform coefficients, and coding the quantized transform coefficients to provide coded transform coefficients. A local decoder expands the quantized transform coefficients to provide an additional prediction picture. A motion detector calculates a motion vector for each of plural subblocks obtained by dividing each block by at least four. A representative motion vector generator generates plural representative vectors from the motion vectors of the subblocks constituting each block. The representative vectors are fewer in number than the number of subblocks per block, and represent the motion vectors of the subblocks. In response to the representative vectors, a motion compensator applies motion compensation to the prediction picture to produce the prediction blocks. In a complementary expander, a demultiplexer separates coded transform coefficients and coded vector data, including the representative vectors, from the compressed picture block. A vector decoder detects and decodes the representative vectors in the coded vector data, decoding fewer representative vectors than the number of subblocks. A calculating circuit calculates the motion vectors of the subblocks from the representative vectors, and a circuit derives a block of the output picture signal from the coded transform coefficients and the motion vectors.
|
