论文部分内容阅读
本文通过对在混合型网络中传输的MPEG4编码视频流进行质量预测，根据用户层的感知质量由下层参数决定的理论建立预测模型，提取不同视频的时、空域特征参数,结合其他应用层参数 qscale, MTU，FR和网络层参数PER，进行多元非线性回归拟合得到反映主观视频质量的参数MOS的预测式，以评价用户感知质量（QoE）。本文提出的视频质量预测模型是非参考的，因此对于在线视频监测有很强的实用性。并且，NS2下的仿真证明，本文提出的预测模型具有良好的准确性。“,”The aim of this paper is quality prediction for MPEG4 video streaming over mixed networks, in order to do this, I found a Cross-layer Subjective quality prediction model to evaluate quality of service (QoE) on subscriber layer. First extract content feature of different video streaming, including temporal and spatial feature information, and then consider other parameters of application layer and network layer. In application layer, we choose encode parameters (Quantitative scale, MTU) and FR, and in network layer choose PER. At last, by multivariate nonlinear regression analysis, we get the prediction equation of MOS and chosen parameters for evaluation on QoE . Because the prediction model of quality of video streaming is reference-free, it can be used to do video observation on line and ensure the quality of video transmission easily and effectively.