SIP视频会议的自适应采集与动态码流提取技术研究

发布时间：2018-04-25 05:16

本文选题：视频会议 + SIP　；参考：《哈尔滨工业大学》2016年硕士论文

【摘要】：视频会议使处于两地的人可以更加方便的交流沟通,提高了人类社会生产的协作效率,在人类社会生产中逐渐占据了重要地位。Internet普及和网络基础设施全面升级,使得基于Internet网络的简单、方便且廉价视频会议系统成为可能。但是在Internet中带宽波动较大,网络通信质量不可控。因此如何保障Internet中视频会议传输QoS(Quality of Service)问题便成一个重要的研究课题。本课题尝试从实时视频码流的码率控制角度来提升视频会议在不稳定的网络中的传输QoS。核心内容是SIP(Session Initiation Protocol)视频会议的基础上加入自适应采集(视频采集)与动态码流(视频码流)提取技术。主要研究内容包括SIP视频会议原理和相关码流技术分析,基于RTP/RTCP反馈机制改进的AIMD视频发送码率控制模型的建立,设计实现SIP视频会议系统。研究分析SIP视频会议协议结构和SIP视频会议视频码流技术。从理论上全面分析一个基于SIP的视频会议的交互流程和架构模型。研究分析H.264视频编码结构和相应码率控制、视频转码方案。根据视频会议场景选择码率控制方案并提出一种基于GOP的全编全解视频转码方案。研究分析基于RTP/RTCP视频传输的网络拥塞控制方案,将传统拥塞控制算法AIMD加以改进应用于自适应视频采集和动态码流提取的码率决策部分。其中利用RTP/RTCP反馈控制机制获取丢包率、回环时间和抖动参数。设计一种双滑动窗口机制用于丢包率的预测。对回环进行低通滤波平滑。根据丢包率、回环时间和抖动参数作为AIMD算法中视频码率决策的参数。在此基础上设计自适应视频采集和动态视频码流提取模型。在OpenMCU、OpenSips等开源项目基础上进行二次开发,设计并实现加入自适应视频采集与动态视频码流提取技术的SIP视频会议系统。主要是在SIP UA端加入自适应视频采集模块,在转发服务器端添加动态视频码流提取模块。该系统的特点是有更高的网络敏感性,能根据网络的差异动态调整系统中传输的视频的码率,能有效的提高视频会议的视频通信质量。本课题充分考虑视频会议的应用场景以及传输码率的预测问题。在基于AIMD算法基础上加以改进,设计出整套能实现自适应视频采集和动态码流提取的网络敏感性视频会议系统,有效提升了系统的QoS。
[Abstract]:Video conference makes it more convenient for people in both places to communicate, improve the efficiency of human social production, and gradually occupy an important position in the production of human society. It makes the Internet network based on the simple, convenient and cheap video conferencing system possible. However, the bandwidth fluctuates greatly in Internet, and the network communication quality is not controllable. Therefore, how to guarantee the QoS(Quality of Service in Internet becomes an important research topic. This paper attempts to improve the QoS of video conferencing in unstable networks from the point of view of bit rate control of real-time video stream. On the basis of SIP(Session Initiation protocol video conference, adaptive acquisition (video capture) and dynamic bit stream (video bitstream) extraction are introduced. The main research contents include the principle of SIP video conference and the analysis of related bitstream technology, the establishment of AIMD video transmission rate control model based on RTP/RTCP feedback mechanism, and the design and implementation of SIP video conference system. The structure of SIP video conference protocol and the video bitstream technology of SIP video conference are studied and analyzed. The interaction flow and architecture model of a video conference based on SIP are analyzed in theory. The H.264 video coding structure and the corresponding rate control, video transcoding scheme are studied and analyzed. According to the video conference scene, the rate control scheme is selected and a fully decomposed video transcoding scheme based on GOP is proposed. The network congestion control scheme based on RTP/RTCP video transmission is studied and analyzed. The traditional congestion control algorithm AIMD is improved and applied to the bit rate decision part of adaptive video capture and dynamic bit stream extraction. The RTP/RTCP feedback control mechanism is used to obtain the packet loss rate, loop return time and jitter parameters. A double sliding window mechanism is designed to predict packet loss rate. The return loop is smoothed by low pass filtering. The packet loss rate, loop time and jitter parameters are used as the parameters of video rate decision in AIMD algorithm. On this basis, the adaptive video capture and dynamic video stream extraction model are designed. On the basis of OpenMCU OpenSips and other open source projects, this paper designs and implements a SIP video conference system with adaptive video capture and dynamic video stream extraction technology. It mainly adds adaptive video capture module in SIP UA terminal and dynamic video stream extraction module in forwarding server. The characteristic of this system is that it has higher network sensitivity, can dynamically adjust the bitrate of video transmitted in the system according to the network difference, and can effectively improve the video communication quality of video conference. In this paper, the application scene of video conference and the prediction of transmission bit rate are fully considered. Based on the improvement of AIMD algorithm, a network sensitive video conference system which can realize adaptive video capture and dynamic bit stream extraction is designed, which can effectively improve the QoS of the system.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN948.63

【相似文献】