Dynamic Parallel Computing Architecture for Video Processing

Abstract

In this paper, motion estimation preprocessing algorithm is mapped onto a new dynamically parallel computing architecture, namely, the parallel computing architecture, which consists of multiple parallel units. It eventually reduces the computation required for motion estimation in advance video coding. A directed acyclic graph is constructed to represent the video coding algorithms comprising motion estimation. This speeds up the video processing with minimum sacrifice.