《计算机应用研究》|Application Research of Computers

一种H.265视频带宽实时预测模型

Real-time prediction model of H.265 video bandwidth

免费全文下载 (已被下载 次)  
获取PDF全文
作者 刘江涛,李泽平,杨炳钊
机构 贵州大学 计算机科学与技术学院,贵阳 550025
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)02-010-0365-05
DOI 10.19734/j.issn.1001-3695.2018.08.0552
摘要 针对视频直播过程中带宽抖动导致视频帧丢失造成用户体验降低问题,提出了一种H.265视频带宽实时预测(VBRP)模型。模型基于马尔可夫链,专注于H.265编码的直播视频流,研究了在GOP(group of pictures)编码模式下B帧出现的统计特性,发现B帧大小对视频流传输速率的影响较大。应用此特性可预测出直播流中B帧的大小,并根据网络带宽选择性丢弃B帧,同时引入步长调整因子AF和误差阈值FT,均衡模型在预测B帧时的训练频率和预测帧数。最后基于模型实现VBRP预测算法并在直播系统中验证算法的有效性。
关键词 带宽预测; H. 265; 马尔可夫链; 流媒体
基金项目 国家自然科学基金资助项目(61462014)
贵州省优秀科技教育人才省长专项资金资助项目(黔省专合字[2011]34)
贵州省科学技术基金资助项目(黔科合J字[2011]2201)
贵州大学引进人才科研项目(贵大人基合字[2010]034)
本文URL http://www.arocmag.com/article/01-2020-02-010.html
英文标题 Real-time prediction model of H.265 video bandwidth
作者英文名 Liu Jiangtao, Li Zeping, Yang Bingzhao
机构英文名 College of Computer Science & Technology,Guizhou University,Guiyang 550025,China
英文摘要 In order to solve the user experience reduction by bandwidth jitter that caused video frame loss during live video, this paper proposed a real-time prediction model(VBRP) of H. 265 video bandwidth. The model was based on Markov chain and focuses on the live video stream encoded by H. 265. It studied the statistical characteristics of frame B in the GOP(group of pictures) encoding mode, which showed that the B frames size had a great influence on the transmission rate of video stream and used this feature, it can predict the size of the B frame. Then according to the network bandwidth selective discarding the B frame. At the same time used the number of frames adjustment factor AF and error threshold FT, to balance the training frequency and the number of predicted frames when predicting B frames. Finally, the prediction algorithm is implemented and validated in the live video system.
英文关键词 bandwidth prediction; H. 265; Markov chain; media streaming
参考文献 查看稿件参考文献
 
收稿日期 2018/8/3
修回日期 2018/9/13
页码 365-369
中图分类号 TP393
文献标志码 A