Toward Optimal Bitrate Allocation in the Scalable HEVC Extension: Application to UHDTV
Résumé
Scalable video encoders compress a single video sequence to produce a bitstream composed of several layers, corresponding to different temporal, spatial and quality representations of the input video sequence. This technique improves the coding efficiency compared to simulcast encoding of each representation by exploiting additional correlations through inter-layer predictions. The latest scalable video coding standard SHVC --the extension of the recent HEVC standard-- announces up to 30% bandwidth reduction. However, this gain is valid under the common test conditions, established by the JCT-VC expert group, which are not necessarily relevant in broadcasting environment and do not include video sequences in UHD resolution. In this paper, we expose the results of an extended study about the optimum gains that scalability can bring in concrete broadcast use-cases. Indeed, we consider SHVC with HDTV/UHDTV as spatial enhancement layers. Then, we search the optimum balance of layers' dedicated bitrates in sense of coding efficiency and objective quality for different UHDTV video sequences and use-cases.