SVC의 복합 스케일러빌리티를 위한실용적인 RTP 패킷화 알고리듬

Abstract

In this paper, we propose a practical RTP packetization algorithm for combined scalability which can be supported by SVC. SVC can support the combination of temporal, spatial, and SNR scalability within one bitstream to provide a wide range of combined scalability. In combined scalability, a coded picture of a base or scalable layer is produced as one or more slices at the video coding layer (VCL). The network abstraction layer (NAL) encapsulates each slice generated by the VCL into typical NAL unit which forms the basic structure of an SVC bitstream. The SVC NAL unit header contains (TL, DID, QL) field to identify the association of each SVC NAL unit with its scalable layer without decoding the SVC bitstream. The temporal_level (TL) indicates the temporal layer hierarchy in temporal scalability. The dependency_id (DID) denotes the inter-layer coding dependency hierarchy in spatial scalability. The quality_level (QL) designates the quality level hierarchy of FGS layer in SNR scalability. In this paper, we utilize the (TL, DID, QL) information to derive hierarchical spatio-temporal relationship of the SVC NAL units in the scalable layers. Based on this derivation, we propose a practical RTP packetization algorithm that can perform efficient packetization for Non-interleaved packetization mode of RTP in combined scalability.