考虑时间-空间特性的流媒体视频QoE评价模型

摘要

流媒体视频服务作为目前网络服务最广泛的应用之一，其用户感知质量已经引起了服务提供商的广泛关注。文章提出了一种预测流媒体视频用户体验质量（QoE）的无参考评价模型。该模型基于人类视觉系统（HVS），考虑视频时间-空间特性，讨论了编码压缩、网络丢包因素对视频质量的影响情况，与用户感知评分（MOS）相关度达到0.965，且二者均方根误差小于0.298，能够较为准确地预测流媒体视频用户感知质量。%In this paper, a no-reference evaluation model was established to predict users' Quality of Experience (QoE)on streaming videos considering video spatial-temporal characteristic based on Hu-man Visual System. Our proposed model was achieved by taking into account both encoding com-pression impairment and network packet loss quality degradation. The Pearson Correlation Coefficient be-tween MOS and Estimated QoE score is as high as 0.965 and the Root Mean Square Error of them is smaller than 0.289. Therefore, the QoE assessment model has a great accuracy on predicting users' quality perception of streaming videos.