Exploitation of Interactive Region of Interest Scalability in Scalable Video Coding by Using an XML-driven Adaptation Framework
Davy De Schrijver, Wesley De Neve, Davy Van Deursen, Sarah De Bruyne, Rik Van de Walle · Ghent University Academic Bibliography (Ghent University) · 2006
Interactive Region Of Interest (IROI) scalability is one of the requirements of the Scalable Video Coding extension of H.264/AVC (SVC). IROI can be used in applications in which the most interesting part of the video pane (in particular, the ROI) cannot be defined during the encoding phase. Therefore, the coded bitstreams have to be created in such a manner that the ROI can be selected on-the-fly at any moment. In this paper, we show how these bitstreams can be generated within the SVC specification by relying on the Flexible Macroblock Ordering (FMO) tool. By using this tool, it is possible to divide a frame in independent decodable tiles or slices. We subsequently explain the extraction process to select the slices belonging to the desired ROI. Furthermore, in this paper, we shift the adaptation process (i.e., the extraction of the ROIs) from the binary domain to the XML domain. This gives us the opportunity to rely on a format-agnostic adaptation engine which only makes use of an XML-based bitstream syntax description and a transformation stylesheet. The use of the MPEG-21 Bitstream Syntax Description Language will be discussed to obtain the XML-based bitstream syntax descriptions. The performance of the different steps in this framework is investigated, in particular, the overhead of coding a bitstream once with and once without the IROI scalability, the generation time of the XML descriptions, the extraction process implemented in Streaming Transformations for XML (STX), the regeneration of the adapted bitstream, and finally the quality and bit rate of the different bitstreams. 1.