【作者】 杨志伟；

【导师】 冯宗哲；

【作者基本信息】 西安电子科技大学 ， 模式识别与智能系统， 2005， 硕士

【摘要】 视频含有海量信息这一特征使得它成为人们获取信息的一个重要来源。对视频信息的利用将不可避免地需要考虑三个方面的问题：如何高质量地获取原始视频信息?怎样稳定可靠地对获取的视频信息进行传输?如何理解这些被包含在原始视频中的信息?简言之：对视频信息的使用将是视频获取、视频传输和视频理解这三个方面的结合。 视频传输是视频信息利用的中间环节，更是展开视频应用的核心环节。本文对视频传输系统进行系统的对比研究，给出了一个针对应用的嵌入式实时视频传输系统的组成框图，并结合所提出的原理框图对各个子模块(视频采集模块、视频编解码模块和传输控制、协议处理模块)进行了详细描述。 在实时视频应用前提下，针对视频编码解码模块指出：对原始视频信息应进行面向传输的视频信源编码，并且为了能够具有对信道传输特性的鲁棒性和最佳的传输效果需要采用联合信源／信道编码、解码技术；为了能够最大限度地应用已经接收的视频数据流，在接收端需要采用错误隐藏和错误恢复技术。 在传输控制、协议处理模块，本文重点集中于网络拥塞控制算法的研究。结合视频流媒体业务需要低传输延时和小的传输延时抖动，并且需要满足与传统数据业务具有对网络资源的公平性的要求，提出了一个面向终端的基于RTT(Round Trip Time)的自适应视频流拥塞控制方法New-AVSP(New Adaptive Video Stream Protocol)。与已有的视频流拥塞控制算法相比：New-AVSP采用基于波形平滑指数和波形突变指数的滑动窗口加权平均RTT估计算法对RTT值进行平滑估计；采用基于网络可用带宽与终端发送速率的参数自适应方法，调整终端注入网络的业务流量。仿真结果表明：New-AVSP较已有的针对流媒体应用的拥塞控制方法而言，在网络传输带宽突变情况下具有更快的逼近速度；在传输带宽小幅度波动情况下具有更为平滑的控制曲线。更多还原

【Abstract】 It is very important to use video as its huge information. For efficiently using video information, we have to face to several problems. How can we get the original video in high visual quality? How to transmit the video information in stable and robust way and how to understand the video information? In other words, video capturing, video communication and video understanding should be integrated for video information utilization.Video communication is not only the inter step but also the more important step in video information application. This paper provides a skeleton for embedded real-time video transmission. Based on the skeleton, the video capturing module, video coding and decoding module, communication controlling and protocol transacting module also is discussed in detail. For efficiently transmit video stream, we point out that the original video information should be encoded for communication and furthermore should make use of joint source and channel encoding and decoding technology for robust transmission and application. We also show that in the receiver error concealment and error resilience should be adopted for maximum using of the rightly received video stream data.The foundation function of the communication controlling and protocol transacting module is how to provide less transmission delay and smooth delay jitter and how to ensure the fairness between video stream and traditional data stream. This paper also introduces a new adaptive video stream congestion control algorithm for end-to-end application (New-AVSP). We define wave flatness parameter and wave variable markedness parameter, and then provide a method for RTT (Round Trip Time) estimation which using flatness parameter and markedness parameter by moving window average way (FMMA). Calculating the available bandwidth at the current time based on RTT and packet loss ration. Then we give an available bandwidth based parameter adaptive moving average algorithm (BPAMA) for send rate control. By contrast with the traditional congestion control method, the simulation results show that the algorithm which we improved has the faster converges rate on the condition of step variety and the smooth ability when wave jitter.更多还原