Study On Scalable Video Coding Based On H.264

Scalable video coding is a new kind of video coding, which can meet the various video services needs, such as internert video, wireless moving video, video monitoring, video phone and real time rebroadcasting. The terminal capability (display size, processing ability, etc) and the transmit channel(band wide, noise and congestion, etc) needed by these different circumstance are always different with each other. The scalable video stream with different rate and different quality is needed. So it is important to carry on the research in scalable video coding.From January 2005, JVT(Joint Video Team), founded by ITU-T and ISO/IEC, began to develop the scalable video coding(SVC) as the new profile of H.264 recommendation. This paper researches and realizes fine granularity scalable(FGS) video coding algorithm in draft version JSVM9.12.2 released by JVT, based on discrete cosine transform(DCT) and bit plane coding. The performance comparison between the proposed fine granularity scalability algorithm and the coarse granularity scalability algorithm in JSVM is performed in this paper.The paper contains five parts as follows:(1) Introduces the development of scalable video coding and its related theories;(2) Analyses the main technologies used by H.264 and the algorithm structure which provides scalability in temporal, spatial and quality domains in JSVM;(3) Discusses the progressive fine granularity scalable video coding algorithm;(4) Develops and realizes the fine scalable video coding algorithm on the base of discrete cosine transform and bit plane coding, performs test for the proposed algorithm in JSVM9.12.2, and makes performance comparision between H.264 and the coarse granularity scalability provided by JSVM.The proposed fine granularity scalable video coding algorithm has low computational complexity. The video stream generated by the algorithm can be truncated arbitrarily at the bit level, and has nice performance in adapting the transmit channel and recoving from error. Because the algorithm is based on discrete cosine transform and bit plane coding, the coding efficiency is lower than common H.264 and the coarse granularity scalable coding in JSVM. So researching new approach to improve the coding efficiency is needed in the future.