CABAC accelerator architectures for video compression in future multimedia: A survey

Abstract

The demands for high quality, real-time performance and multi-format video support in consumer multimedia products are ever increasing. In particular, the future multimedia systems require efficient video coding algorithms and corresponding adaptive high-performance computational platforms. The H.264/AVC video coding algorithms provide high enough compression efficiency to be utilized in these systems, and multimedia processors are able to provide the required adaptability, but the algorithms complexity demands for more efficient computing platforms. Heterogeneous (re-)configurable systems composed of multimedia processors and hardware accelerators constitute the main part of such platforms. In this paper, we survey the hardware accelerator architectures for Context-based Adaptive Binary Arithmetic Coding (CABAC) of Main and High profiles of H.264/AVC. The purpose of the survey is to deliver a critical insight in the proposed solutions, and this way facilitate further research on accelerator architectures, architecture development methods and supporting EDA tools. The architectures are analyzed, classified and compared based on the core hardware acceleration concepts, algorithmic characteristics, video resolution support and performance parameters, and some promising design directions are discussed. The comparative analysis shows that the parallel pipeline accelerator architecture seems to be the most promising. © 2009 Springer Berlin Heidelberg.