《交互式多视点视频系统的编码及相关技术研究》由会员分享，可在线阅读，更多相关《交互式多视点视频系统的编码及相关技术研究（91页珍藏版）》请在金锄头文库上搜索。
1、宁波大学 硕士学位论文 交互式多视点视频系统的编码及相关技术研究 姓名：Renlong He 申请学位级别：硕士 专业： 指导教师：蒋刚毅 20090105 宁波大学硕士学位论文 III 交互式多视点视频系统的编码及相关技术研究 摘 要 随着计算机、通信及网络技术的发展，交互式多视点视频系统的应用前景已 经越来越广泛，其应用包括立体电视、远程视频会议等等。本文以交互式多视点 视频系统框架为背景，研究其中的编码及相关技术。 首先，本文基于多视点视频序列相关性，提出独立片的概念，进而提出一种 采用自适应参考帧模式选择的多视点视频编码方法。在保证高率失真性能的前提 下，应用独立片、自适应算法有效地降
2、低了多视点视频编码的计算复杂度，将多 种性能进行有机结合，可以根据用户外部设置的参数自适应调整参考帧模式。实 验结果表明，该方法在保证高压缩效率的前提下，能有效地降低计算复杂度，并 改善随机访问性能。 其次，由于深度图压缩处理之前，需要进行量化处理。而采用哪种量化方式， 相关研究人员目前仍未达成一致观点。本文通过实验来研究两种量化方法对深度 图编码、绘制的影响，发现采用均匀量化方式时，深度图的编码效率及绘制质量 均高于非均匀量化情况。 接着，基于深度图中边缘部分对最终虚拟视点绘制效果影响最大这个特点， 本文提出了一种应用于 3DTV(Three Dimensional Television)系
3、统的基于边缘增强的深 度图编码压缩算法。它在保证最终绘制效果不变的前提下，采用边缘提取、自适 应分片等相关处理过程，对不同的深度图区域使用了不同的量化系数。实验结果 表明，所提出的深度图编码压缩算法在最终绘制效果不变的前提下，较大程度地 节省了码率，进而减轻了网络带宽压力。 最后，为了提升多视点视频系统中随机访问的性能，本文提出两种算法。1) 提出帧等级矩阵的概念，进而给出针对 HBP(Hierarchical B Predictive)这种具有代表 性的多视点编码结构的解码路径计算快速算法； 2) 将多视点视频编码结构抽象为 加权图形式，并引入图论中的最短路径思想，从图模型分析的角度提出了一
4、个多 视点视频系统中随机访问解码路径计算改进算法，以实现快速随机访问。实验结 果表明，所提算法能够快速正确地计算出最短解码路径，从而有效地减少随机切 换响应时间，降低多视点视频流传输的带宽占有率。 关键词：交互式多视点视频系统，深度图区域，虚拟视点绘制，随机访问 宁波大学硕士学位论文 IV Coding and related Technology Research in Interactive Multi-view Video System Abstract As the development of computer, communication and network technolog
5、y, the application prospect of interactive multi-view video system is more and more broad. Its application including 3DTV(Three Dimensional Television), remote video conferencing and so on. The study subject of this thesis is the research on coding and other related techniques, which is based on the
6、 architecture of interactive multi-view video system. Firstly, based on the correlations of multi-view video, the concept of independent slice is introduced. And a multi-view video coding method with adaptive selection of reference frame modes is proposed. The introduction of independent slice and a
7、daptive selection of reference frame modes greatly contribute to the increase of coding efficiency as well as the reduction of computational complexity. In addition, the proposed method combines many MVC(Multi-view Coding) prediction schemes and is able to dynamically select suitable coding modes ac
8、cording to the users decision. Experimental results show that the proposed multi-view video coding method can effectively reduce the computational complexity and improve the random access performance while maintaining high coding efficiency. Next, in 3DTV system, virtual view generation can be accom
9、plished by DIBR (Depth Image-based Rendering) system. And, depth map is key side information for generating virtual image by using DIBR technique. Before compressing, depth map need to be quantified. But, which quantification method should be choose or how to transform depth information is still a c
10、ontentious issue. In this paper, we compare the effects of two depth map quantification methods for coding and synthesizing. By the experiments, we show that the depth map is coded efficiently in terms of the bit rate and quality of the synthesized image when it uses uniform quantization method. The
11、n, a depth map coding method is proposed for 3DTV system in this paper. The proposed method is based on edge enhancement algorithm by considering the fact that the quality of synthesized virtual viewpoint images will be greatly affected by boundary areas in depth map. Using the process of edge extra
12、ction and slice-adaptive, the proposed method uses different quantitative coefficients for different depth map regions while maintaining quality of synthesized virtual viewpoint images. Experimental results indicate that the proposed depth map coding scheme saves bit-rate markedly while maintaining
13、quality of synthesized virtual viewpoint images. Finally, in order to improve random access performance in multi-view video system, this paper proposed two algorithms. 1) A concept of frame level matrix is proposed for decoding path computation, and then a fast decoding path computation algorithm is
14、 proposed for the H.264-based 宁波大学硕士学位论文 V multi-view video coding with the prediction structure of HBP(Hierarchical B Predictive); 2) Abstracts the Multi-View coding structure to the weighted map model. With analysis of the corresponding graph model, a decoding path computation method is proposed t
15、o obtain fast random access, based on the shortest path idea. Experimental results show that the proposed method can obtain the shortest decoding path, so it effectively reduces the response time of random access and bandwidth occupancy rate in multi-view video system. Key Words: Interactive Multi-v
16、iew Video System, Depth Map Region, Virtual View Synthesis, Random Access 宁波大学硕士学位论文 II 独独 创创 性性 声声 明明 本人郑重声明：所呈交的论文是我个人在导师指导下进行的研究工作 及取得研究成果。尽我所知，除了文中特别加以标注和致谢的地方外，论 文中不包含其他人已经发表或撰写的研究成果，也不包含为获得宁波大学 或其他教育机构的学位或证书所使用过的材料。与我一同工作的同志对本 研究所做的任何贡献均已在论文中做了明确的说明并表示了谢意。 签名：_ 日期：_ 关于论文使用授权的声明关于论文使用授权的声明 本人完全了解宁波大学有关保留、使用学位论文的规定，即：学校有 权保留送交论文的复印件，允许论文被查阅和借阅；学校可以公布论文的 全部或部分内容，可以采用影印、缩印或其他复制手
