Research On Scalable Video Coding For Transmitting In Internet

With the further developments of the Internet, besides the original Internet applications such as E-mail, Telnet, FTP etc, the application of media stream communication in the Internet, including video conference, video on demand (VOD), distance education, become a very important kind of network communication application, also is a research focus. Because IP network has some inherent characters such as delay, transmit error, packet loss and bandwidth fluctuation, it can only afford best-effort service, but cannot offer Quality of Service (QoS) guarantees for the real time video stream transmission, which is a great challenge to video compression coding. On the other hand, the heterogeneity of the networks and the variety of terminal equipments make it hard to transmit video stream generated by traditional storage-based video compression coding, which tries to improve the coding efficiency and get the optimal video quality in a fixed bit-rate and is not adapt to transmit in the Internet because it cannot work in an efficient and flexible way according to the bandwidth utility and QoS guarantee. So it is necessary to find a new video compression coding and transmitting method that is suitable for the IP network and network-based scalable video coding meet the requirment to some extent.Firstly, we summarize the researches and developments of media stream service and introduce some important technologies and relevant international standards in video coding. Secondly, we analyse the basic temporal scalability, spatial scalability and SNR scalability introduced in MPEG-2 and MPEG-4. Then we research in detail on the fine granularity scalability (FGS) defined in MPEG-4 AMD4, including the integration of FGS with the temporal, spatial scalability. In regard to the low coding efficiency of FGS, we present an improved FGS. Comparing to the fixed bit-rate allocation for the base layer in FGS, the bit-rate allocation algorithm in the improved FGS tries to allocate the bit-rates for the base layer and the enhanced layers adaptively according to the various network conditions. This algorithm improves the video quality by increasing the quality of the reconstruction picture of the base layer and the precision of movement estimate. In our own test video coding scheme, the experimental results shows that the algorithm improve the coding efficiency about 0.5dB in the low and middle bit-rate range. Thirdly, we analyse and implement the progressive fine granularity scalability (PFGS) and further discuss the macroblock-based PFGS. In order to improve the resilience and robust of PFGS enhanced layer bit-stream, we introduce some resilient algorithms used in MPEG-4 to PFGS enhanced layer coding. We also use Universal VLC (UVLC) in the coding to limit the transmit error in a few blocks because of its excellent self-synchronization property. At last, we present some proposals in scalable coding.