【作者】 刘艳；

【导师】 尔桂花；

【作者基本信息】 清华大学 ， 控制科学与工程， 2004， 硕士

【摘要】 随着Internet和无线通信的飞速发展,人们在网络上实时的获取图像和视频数据已经成为可能,相应的流媒体技术也应运而生。然而由于网络的异构性、数据丢失率、传输延迟等的存在,给视频编码技术提出了新的挑战。传统的图像和视频的编码也正从面向存储向面向网络转变,此时可分级性编码及其网络传输问题得到了广泛的重视。本论文主要研究了可分级编码,特别是面向网络的精细可扩展编码,以其在网络上的相关传输问题。首先,论文研究了可分级编码方法及其最近发展的成果。内容包括时域、空域、质量、混合可分级在内的可分级编码的基本模式,重点分析和研究了近年提出的面向网络的精细可扩展性编码(FGS)和渐进的精细可扩展性编码(PFGS),以及基于小波的编码等。其次,精细可分级编码的码率分配问题得以解决。论文总结了FGS率失真函数的观察特性,提出了快速码率分配的解决方案,该方案在保证视频质量平滑性的同时,能够大大提高码率分配的效率,满足实时性的需求。针对突发错误信道上进行固定时延传输的码率分配问题,本文设计的“迭代等D优化”算法在理论上能够达到相应码率问题的最优解,仿真实验也表明该算法具有相对其他算法更优的视频传输质量和效果。最后,论文对精细可分级码流传输系统进行了实现和研究。系统建立了一个较完整的精细可分级编码的视频点播平台,实现了流媒体传输的核心协议,对系统的服务质量问题提供了解决方案,为各种视频网络传输技术的研究提供了实验平台。在此流媒体传输平台的基础上,提出了基于缓冲区的自适应传输FGS码流的方案,方案基于缓冲区启动策略,结合了滑动窗口的思想实现了FGS码流的自适应传输,性能的分析和实验显示了这种方案能够达到自适应网络带宽和平滑视频质量的传输效果。更多还原

【Abstract】 With the development of Internet and wireless, it is easier for people to obtain real-time image and video over networks, which generates the technology of streaming media. However the characteristics of Internet, such as heterogeneity, packet loss, delay, also give new challenges to the research of video coding. Then the goal of video and image coding is changing from storage oriented to network oriented. And the scalable coding is largely applied and appreciated in the field of video transmitting over Internet. And this paper mainly researches on scalable coding, especially Fine Granular Scalability(FGS) coding, and the problems of its transmission. Major work and contribution could be described as follows.Firstly, this paper researches on scalable coding methods. The research includes the basic modes of scalability, which include SNR, spatial, temporal, and hybrid scalability, and FGS and PFGS(Progressive Fine Granular Scalability) which are proposed recently. Wavelet Coding and its research approach are also discussed for its natural scalability. Secondly, rate allocation problem of FGS bit-stream is discussed and solved. An observation characteristic of FGS Rate-Distortion(R-D) relation is proposed and discussed. Then a fast rate allocation scheme is presented, which largely enhances the efficiency of rate allocation and smoothes the video quality at the same time. For the rate allocation problem for delay constrained video transmission over burst-error channels, the paper designs an algorithm named “Iterative Equi-Distortion Optimization ”(IEDO), which is demonstrated to get the optimized solution for the corresponding problem and get better video quality than other algorithms.Thirdly, we establish a FGS transmission platform and do some researches on it. The platform provides almost complete functions of a VOD <WP=5>system. It realizes kernel protocols of a streaming media system and also gives solutions for Qos problems over Internet. And it provides a good platform for study in video streaming problems. Based on this platform, a scheme of adaptive FGS transmission is proposed and studied. The scheme is driven by sender and receiver buffer, and also utilizes the idea of sliding windows to provide adaptation, which is demonstrated to transmit FGS video adaptively and smoothly.更多还原