Method and apparatus for calculating a pixel block from one or two prediction blocks

摘要

A method is disclosed for calculating the pixel values of a sub-pixel accuracy motion compensated block of video pixels from one or two reference blocks, as is required in typical digital video compression and decompression systems. This method uses a minimal amount of temporary storage memory resulting in a compact architecture suited for inexpensive consumer applications. This invention utilizes a pixel pipeline within a block line pipeline to calculate the half pel accurate reference blocks and to average two blocks to result in a prediction block of pixels. The lines from each reference block are input to the invention alternately resulting in reduced memory requirements.