| 摘要 |
Apparatus for generating a motion vector for use in compressing a digital motion picture signal. The apparatus comprises a circuit for obtaining a primary motion vector with one pixel precision between a picture block of the current picture and a designated block of the reference picture. The apparatus also includes a parallelizing circuit for convening the pixel data of the reference picture into parallel pixel data. The apparatus also includes a search area selecting circuit for selecting the parallel pixel data of a search area of the reference picture in response to the primary motion vector. The search area surrounds and includes the designated block. An interpolating circuit generates parallel half-pixel data from the parallel pixel data of the search area by interpolation in plural parallel paths. The interpolating circuit also selects sets of parallel half-pixel data from the parallel half-pixel data generated in the plural parallel paths, and serializes the sets of parallel half-pixel data to provide parallel matching blocks of half-pixel data. The matching blocks have shifts relative to the designated block of -1/2, 0, and +1/2 pixel, horizontally and vertically. The apparatus further includes a circuit for identifying the one of the matching blocks that most closely matches the picture block as a reference block. Finally, a motion vector calculating circuit calculates the motion vector with half-pixel precision from the primary motion vector and the shift of the reference block.
|
