Efficient Transformation of MPEG-21 Metadata for Codec-agnostic Adaptation in Real-Time Streaming Scenarios

Abstract

Scalable media contents, such as the new MPEG-4 scalable video codec enable to easily retrieve different qualities of the media content by simply disregarding certain media segments. The MPEG-21-based codec-agnostic adaptation approach supports this concept by introducing anXML-based bitstream syntax description (BSD) which describes the different segments of a media content. Based on this BSD, an adaptation node can intelligently adapt any scalable media (i.e., remove specific media segments) without the need for codec-specific knowledge. The adaptation approach consists of 1) transforming this BSD and 2) adapting the media based on the transformed BSD. In this paper, we focus on the BSD transformation step and evaluate different mechanisms w.r.t. their transformation efficiency given several application scenarios. In particular, we compare the traditional style sheet-based mechanisms with a novel mechanism based on regular expressions. We discuss both mechanisms in terms of their expressiveness, and propose how to actually employ regular expressions for codec-agnostic adaptation. Finally, we quantitatively evaluate these mechanisms in different adaptation scenarios, which vary in the size and number of required BSD units.