摘要
针对实时多媒体在ＩＰ网络上的传输，提出了一种基于实时传输协议的ＭＰＥＧ ＶＢＲ视频源的建模，此模型考虑了场景的变化和同一场景内码流的波动。用几何分布来描述场景的变化，用ＡＲ（２）模型来反映同一场景内码流的波动。对互帧采用合成模型建模，对Ｐ，Ｂ帧采用伽玛函数建模。通过对它们的排队性能的分析，证明了这种合成模型能够很好地反映实际视频流的特征。
Considering the transportation of real-time multimedia on IP networks, a model of RTP-based\lPEG VBR video sources is provided to capture the frame size at multiple time scales: the change of scene and fluctuation of bit-rate. Geometric distribution is used to depict the change of scene while the fluctuation is described by the AR(2) process in the same scene.A resultant model is used to model the I frames while the P,B frames are modeled with Gamma density function. The analysis of queue performance proves that the model is good enough to model the MPEG video sources traffic.
出处
《信号处理》
CSCD
2002年第2期180-182,166,共4页
Journal of Signal Processing
