System and method for multi-row decoding of video with dependent rows

摘要

A system and method for decoding digital video by processing multiple regions of an image in parallel, even when there are dependencies between rows in the image, are disclosed. The method generally involves decoding multiple rows concurrently, with the start of decoding of a given row being delayed until portions of the other rows on which the given row depends have been decoded. The system generally comprises parallel processors, with one processor typically decoding one row and another processor typically decoding the row above it. In accordance with the present invention, however, any number or type of processors can decode, or perform decoding functions on, the image in parallel.

主权项

1. A system for decoding, in parallel, more than one row of video data of a picture, the system comprising:
a first processor that performs a decoding function on blocks of pixels of a first row of video data of the picture, wherein each row of the picture cannot be decoded independently of other rows in the same picture; and a second processor that performs the decoding function on blocks of pixels of a second row of video data of the same picture; wherein the second processor begins performing the decoding function on a block of pixels of the second row only after the first processor performs the decoding function on each of a plurality of predecessor blocks of pixels of the first row upon which decoding of the block of pixels of the second row individually depends, wherein the decoding function comprises one of a complete decoding of a block of pixels or a performance of one or more stages of a decoding pipeline.