【作者】 王樟；

【导师】 严国萍； 柳健；

【作者基本信息】 华中科技大学 ， 信息与通信工程， 2007， 博士

【摘要】 近年来,随着互联网和视频编码技术的发展,网络视频应用越来越影响人们的生活方式。例如,视频会议、视频点播(Video-on-Demand, VoD)、视频监控等应用。然而要在互联网上传输视频数据,对视频编码算法,网络传输技术等方面又提出了诸多挑战。其一,由于网络自身的异构性和服务质量(Quality-of-Service, QoS)的不确定性,使得带宽波动和误码、丢包等现象经常发生;其二,由于不同的视频应用对视频传输的质量、时延、可交互性等要求也不同,所以要求传输的视频码流在服务质量QoS方面具有灵活的可适应性;其三,不同的客户端具有不同的带宽接入能力和视频解码播放能力,这就要求传输的视频码流具有相应的自适应结构以方便客户端的数据接收和解码;其四,在视频数据传输过程中,需要根据当前网络健康状况和应用服务的特定性要求,选择最方便快捷的数据传输技术,以保证视频传输的实时性和可靠性。所有上述问题的核心在于要求视频编码和网络传输能结合在一起,并在一定网络条件下最大限度的满足客户端多种多样的应用需求。最新的以H.264/AVC(Advanced Video Coding)为基础的可伸缩性视频编码标准(Scalable Video Coding, SVC)和基于点对点(Peer-to-Peer, P2P)技术的视频传输机制就是这样的两项非常有希望的技术。本文以网络视频应用系统为研究框架,同时从系统中的视频编码模块和视频传输模块入手,较深入地展开两方面研究:1)支持空域可伸缩性的视频编码相关算法,为降低层间上采样滤波运算量,研究其中的层间上采样滤波方案,为进一步拓展层间预测技术,研究适用于增强层上的扩展帧内预测算法;2)基于P2P技术的视频流传输算法,针对当前在非对称速率信道上进行P2P视频传输时所存在的实际问题,研究在多服务节点、单请求节点的传输模式下基于P2P的视频流快速传输算法。具体的研究工作和内容如下:从网络视频应用系统的角度出发,对系统组成结构作了功能模块划分。详细阐述了系统中关于视频编码和视频传输模块的理论基础,包括传统的视频编码技术、可伸缩视频编码技术、SVC标准中的关键技术以及P2P视频传输网络、协议和性能参数。研究了空间可伸缩视频编码的上采样滤波方案。在总结当前图象插值算法研究现状的基础上,提出了一种简单高效的区别图象分量的上采样滤波方案。该方案结合人眼视觉系统的基本特性,综合考虑采样精度和滤波运算量两方面因素,设计了三种不同复杂度的上采样滤波器,分别用于图象中不同分量的上采样滤波。通过对所有测试图象序列的实验结果表明,该方案均能够在保持良好的编码性能基本不变的前提下,大大降低滤波运算量。研究了空间可伸缩视频编码的帧内预测编码算法。在总结当前帧内预测算法研究现状的基础上,结合可伸缩性视频编码标准SVC中的层间预测原理,提出了一种适用于空间增强层的扩展帧内预测算法。新算法以H.264/AVC中的帧内预测算法为基础加以扩展,自适应的在本层邻块数据和基本层对应块解码重构数据之间选择最佳的数据组合作为预测数据源,用来预测编码当前宏块。特别是在本层邻块数据无法获取的情况下,利用基本层解码重构数据作为扩展预测源,使得帧内预测的预测精度明显提高。测试实验结果表明,该算法能提高编码性能,同时没有额外增加编码的计算复杂度。研究了基于P2P的视频流数据传输算法。在总结当前基于P2P的视频传输网络研究现状的基础上,分析了P2P视频传输技术和非对称速率信道的各自特点,建立了一个传输模型,用于描述服务节点个数、上传速率与节点下载速率之间的关系。然后针对当前模型中下行链路带宽利用率低下这一问题,提出了一种基于P2P的多点视频流快速传输算法。该算法将多个服务节点的视频数据经过组合排序后并行上传到请求节点,从而加快下载速率,提高下行链路的带宽利用率。仿真实验结果表明,该算法能够提高网络带宽利用率,同时节省视频流数据传输时间。本文的研究工作同时覆盖了视频编码和视频传输领域中的相关技术内容,具有一定的理论研究意义和实际应用价值。更多还原

【Abstract】 With the development of both Internet and video coding techniques, video applications over Internet have transformed our lives through the way we communicate, such as video conference, video on demand (VoD), video monitoring service and so on. On the other hand, it will face some new challenges on video compression and network transmission over Internet. Firstly, the fluctuatation of bandwidth, packet loss and code error happen frequently due to heterogeneous networks and uncertain quality of service (QoS). Secondly, video compression for streaming should be adaptive on QoS to satisfy people’s different requirements on quality, delay and interaction. Thirdly, video compression should support different decoder capacity and different requirements of play back functionality; it is because different clients have various network access capacity and video decoding capacity. Finally, according to the current status of network healthy and special requirement of application service, the best transmission mode should be chosen to guarantee the requirements of delay and reliability during the video data transmission.All these factors require the cooperation of video compression and network transmission to satisfy different needs of streaming applications. The newest scalable video coding (SVC) scheme based on H.264/AVC and the P2P-based video transmission techniques are most two promising ones. In this thesis, we investigate some new algorithms of these two techniques: 1) spatially scalable video coding algorithms, in order to improve prediction precision, investigate the upsampling filter design, in order to extend the inter layer prediction technique, study the intra coding method for spatial enhancement layer. 2) The P2P-based video transmission scheme. In order to handle the problem of P2P video transmission in the asymmetrical speed channel, study the fast video transmission algorithm in the case of multiple service nodes to single request node. The main contributions of this thesis are as follows:First, the basic theoretical knowledge of video coding and video transmission is deeply introduced and described, which involves traditional video coding, scalable video coding techniques and video transmission networks, protocols and performance parameters. Second, at the aspect of spatially scalable video coding, a simple and efficient image component adaptive upsampling filter scheme in SVC is proposed. In order to increase coding efficiency of SVC, proposed scheme utilizes the basic characteristic of human vision system and differentiates image components. Considering both computational complexity and interpolation precision, it assigns the most suitable upsampling filters for interpolating different components. Experimental results of all testing video sequences show that proposed scheme causes negligible decrease on coding performance and reduces a lot of computational complexity.Third, at the aspect of inter layer prediction technique in SVC, an efficient enhancement layer scalable intra coding algorithm for spatial scalability in SVC is proposed. According to the coordinate attribute of the macroblock to be coded, proposed algorithm makes use of both enhancement layer data and base layer reconstructed data, and chooses the best data combination for intra prediction. Especially, at the case of that the enhancement layer data is not available, base layer reconstructed data is used as predictor instead for intra prediction, which extends the implementation of intra coding under this special situation. Experimental results show that some visible improvement of coding performance is achieved with almost no changes on computational complexity.Finally, at the aspect of bidirectional speed mismatch faced by video transmission in the asymmetrical network environment, a point to multi-point based dummy symmetrical speed video transmission algorithm is proposed. It groups up the video data packets from several source nodes and simultaneously sends them to the destination node, which makes the bidirectional speeds match approximately. Simulation results show that proposed algorithm can increase the utilization ratio of network bandwidth and also reduce the transmission past time of video data packet.The research work involves the techniques in both video coding and transmission domains; therefore, it has important theoretic and practical significances.更多还原