MPEG reconfigurable video representation

Abstract

The standardization efforts of MPEG in video coding, originally had as main -objective to guarantee interoperability of compression systems. This carried with it the possibility to reach another important objective, namely to wide and easy deployment of implementations of those standards. While at the beginning MPEG-1 and MPEG-2 were only specified by textual descriptions, with the increasing complexity of video coding tools, starting with the MPEG-4 set of standards, C or C++ specifications, called reference software, have also become a formal specification of the standards. However, descriptions composed of non-optimized non-modular software packages have shown limitations. Since in practice they are frequently the starting point of an implementation, system designers must rewrite these software packages not only to try to optimize performance, but also to transform such specifications into appropriate forms adapted to be the starting point of current system design flows.