面向DASH的三维和可伸缩视频编码技术研究

发布时间：2018-09-02 09:54

【摘要】：随着互联网和计算机技术的发展,在互联网上进行多媒体数据的传输变得流行起来。随着最新一代的视频压缩编码标准高性能视频编码(HEVC)的发布,基于它的三维视频编码和可伸缩视频编码相关标准也被制定出来。与此同时,为了更好地在互联网上传输流媒体,运动图像专家组(MPEG)开发了基于超文本传输协议的动态自适应流媒体(DASH)标准。传统的DASH系统中使用两维视频编码,对于可伸缩视频编码的研究和使用也只限于上一代可伸缩视频编码标准SVC。由于基于高性能视频编码标准的三维和可伸缩视频编码标准刚刚制定完成不久,将它们用于DASH中的相关研究也比较少。为此,本文面向DASH应用,研究三维视频和可伸缩视频编码,显得非常有意义。具体地,本论文的工作和创新之处主要包括以下方面：首先,本文提出了一个深度图像灵活分辨率编码框架。本文同时还考虑了深度图像的裁切窗口大小和位置,以及深度图像和纹理图像采样窗格之间的偏移。在编码时,对深度图像采用改进的考虑合成失真的率失真优化准则,使得将灵活分辨率编码的深度图像用于虚拟视角合成时能够保证良好的合成视频质量。接着,本文提出了两种提升SHVC下行码流率失真性能的方法。本文分析了采用可伸缩视频的DASH系统中下行码率升高的原因,并针对性地利用SHVC的时间分层级预测结构和层智能启动技术,提出了基于时间子层选择的层间预测和采用开放图像组结构的增强层方法。相比于其它方法,本文提出的这两种方法在有效地提升客户端下行码流率失真性能的同时,能够做到标准兼容。最后,本文提出了一个基于DASH性能指标参数的DASH系统。和传统的基于二维视频编码标准的DASH系统不同,采用三维视频编码以及可伸缩视频编码的DASH系统的编码结构复杂得多。本文针对三维和可伸缩视频编码,提出了压缩性能、接收性能和切换性能三大性能参数,使得DASH系统中的编码端、服务端、代理和缓存设备以及客户端依据相应的指标参数进行编码、缓存、接收和切换。仿真实验表明,采用本文提出的系统框架,能够有效地提升编码、接收和切换性能。
[Abstract]:With the development of the Internet and computer technology, multimedia data transmission over the Internet has become popular. With the release of the latest generation of video compression coding standard, High Performance Video Coding (HEVC), the related standards of 3D video coding and scalable video coding based on HEVC have been formulated. In order to transmit streaming media over the Internet, the Motion Picture Expert Group (MPEG) has developed a Dynamic Adaptive Streaming Media (DASH) standard based on Hypertext Transfer Protocol (HTTP). Traditional DASH systems use two-dimensional video coding, and the research and use of scalable video coding is limited to the previous generation of scalable video coding standard SVC. Three-dimensional and scalable video coding standards that can be used in video coding standards have just been formulated, and there are few related researches on them in DASH. Therefore, it is very meaningful to study three-dimensional and scalable video coding for DASH applications. Firstly, a flexible resolution coding framework for depth images is proposed. The size and position of the clipping window of the depth image and the offset between the depth image and the texture image sampling pane are also considered. In the coding process, an improved rate-distortion optimization criterion considering synthetic distortion is used to optimize the depth image. Secondly, this paper proposes two methods to improve the rate distortion performance of the downlink bit stream of SHVC. This paper analyzes the reason why the downlink bit rate of the DASH system using scalable video increases, and makes use of the time hierarchy of SHVC. Predictive structure and layer intelligent start-up technology are proposed. Interlayer prediction based on time sublayer selection and enhancement layer method based on open image group structure are proposed. Compared with other methods, the two methods proposed in this paper can effectively improve the rate distortion performance of downlink bit stream at the same time, and achieve standard compatibility. A DASH system based on performance parameters of DASH is proposed. Different from the traditional DASH system based on two-dimensional video coding standard, the encoding structure of DASH system based on three-dimensional video coding and scalable video coding is much more complex. Large performance parameters make the encoder, server, proxy and cache devices and client in DASH system encode, cache, receive and switch according to the corresponding index parameters.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN919.81

【相似文献】