Dynamic full-scalability -conversion in scalable video coding

Abstract

Currently, the Scalable Video Coding (SVC) is being standardized. With scalability of SVC, QoS adaptive video streaming service is enabled in heterogeneous network using only one original bitstream. But current SVC lacks dynamic conversion of scalabilities, thereby the adaptation of bitrate to fluctuating network is restricted. In this thesis, we propose dynamic full-scalability conversion method for QoS adaptive video streaming in H.264/AVC SVC. To accomplish the objective, we propose bitstream extraction, encoding and decoding schemes. In encoder, we insert IDR NAL unit to solve the problem of spatial scalability conversion. In extractor, we analyze the SVC bitstream to get the information which enables dynamic extraction. By using this information, we achieve the real time extraction. Finally, in decoder, we define the process to manage the spatial scalability change. The experimental results show that dynamic full-scalability conversion is necessary for time varying network condition.