HEVC与3D-HEVC快速编码算法的研究

发布时间：2018-03-09 04:19

本文选题：HEVC　切入点：3D-HEVC　出处：《吉林大学》2017年博士论文　论文类型：学位论文

【摘要】：随着显示技术的进步,视频已经逐步迈入高清、超高清时代,但是清晰度的提升伴随着数据量的大幅增加,这给视频的存储与网络传输带来负担,现行的视频编码标准H.264/AVC已很难满足视频编码的压缩率要求,新一代视频编码标准High Efficiency Video Coding(HEVC)应运而生,除此之外,因近年来3D技术的进步,出现了众多新型3D显示技术,例如,多视点显示技术,自由视点显示技术等等,伴随着这些新兴技术而来的是更大的数据量,因此3D视频编码也成为了研究热点,而HEVC编码标准也在3D编码方面进行了扩展,提出了3D-HEVC编码标准。HEVC通过加入众多新编码工具使得编码压缩率相对于H.264/AVC提升了一倍,同时保持了视频的质量,代价是大幅增加了编码的复杂度。而3D-HEVC更是加入了深度图编码技术,使得这种复杂度成倍增长,使HEVC以及3D-HEVC的应用受到阻碍,因此降低HEVC以及3D-HEVC编码复杂度,优化编码算法至关重要。本文首先对视频编码技术及其3D扩展进行介绍,简要概述视频编码标准的发展历史,然后对HEVC预测编码的结构框架进行分析,着重对影响编码复杂度的相关编码工具进行介绍,并简要列举视频快速编码算法中的热点。针对HEVC以及3D-HEVC视频预测编码中编码单元尺寸选择过程,以及编码单元预测模式选择过程中计算复杂度过高的问题,提出了如下三种快速算法:1.提出一种HEVC快速帧间编码算法,利用视频图像边缘信息与时空间相邻编码单元的编码信息来降低HEVC编码复杂度,该算法包括了一种快速全2N×2N模式选择算法,一种快速编码单元深度层次划分算法和一种快速预测单元模式选择算法。通过边缘信息来表达编码单元的结构复杂度,利用当前编码单元与时空间相邻编码单元之间的边缘信息差值来反映编码单元内的边缘相似性和边缘移动性,并将边缘相似性和边缘移动性与时空间相邻编码单元的深度层次信息与预测单元的率失真代价值相结合,实现编码单元分割形式以及预测单元模式的提前判断。实验结果表明,这一快速帧间编码算法,在随机接入(RA)模式和低延迟B帧(LD-B)模式下,在与HEVC原编码算法的对比中,分别提升平均53.7%和54.9%的编码速度。2.提出一种基于视频内容的HEVC快速编码算法,对视频内容进行分析,得出3种能够反映视频内容的参数,分别是编码单元内像素的均值、方差以及边缘梯度幅值,并与相邻编码单元的编码信息相结合,实现快速编码单元尺寸选择以及快速编码单元模式选择。实验结果表明,所提快速算法可以在RA模式和LD-B模式下,较HEVC原编码算法分别提升平均58.9%和57.6%的编码速度。3.提出了一种基于深度图的3D-HEVC快速编码算法,来降低纹理视频的编码复杂度。在这一快速算法中我们将深度图编码单元中不同深度值的个数以及深度值的均方误差作为两个参数,并通过这两个参数实现编码单元尺寸的提前决定以及Skip模式的提前判断。最后总结本文的各项研究成果,并预测了视频编码领域未来的研究方向。
[Abstract]:With the development of display technology, video has gradually entered the era of Ultra HD, HD, but enhance the clarity of data with a substantial increase in volume, the burden of storage and network transmission of video, H.264/AVC video encoding standards have been difficult to meet the requirements of video encoding compression ratio, a new generation of video encoding standard High Efficiency Video Coding (HEVC) came into being, in addition, due to the progress of 3D technology, there are many new 3D display technology, for example, multi view display technology, display technology and free viewpoint, accompanies these emerging technologies is the larger amount of data, so 3D video encoding has also become a hot topic however, the HEVC encoding standard in 3D encoding is extended, proposed the 3D-HEVC encoding standard.HEVC by adding many new encoding tools makes the encoding compression rate relative to H.264/AVC upgrade One times, while maintaining the video quality, the price is a substantial increase in the complexity of encoding. While 3D-HEVC is added to the depth map encoding technology, the complexity of exponential growth, the HEVC and 3D-HEVC application is blocked, thus reducing HEVC and 3D-HEVC encoding complexity, optimization is crucial in this paper encoding algorithm. Introduction of video encoding technology and 3D expansion, a brief overview of development history of video encoding standard, then the HEVC prediction analysis frame encoding, encoding tools focus on the impact of the complexity of encoding is introduced, and a brief list of hot video fast encoding algorithm. The HEVC and 3D-HEVC video encoding in the encoding unit size prediction the selection process, calculation complexity and encoding unit prediction mode selection process, put forward the following three algorithms proposed: 1. A fast inter frame encoding algorithm HEVC, to reduce the complexity of HEVC encoding the encoding information of video image edge information and spatial adjacent encoding unit, the algorithm includes a fast 2N * 2N mode selection algorithm, a fast encoding unit depth partitioning algorithm and a fast prediction mode selection algorithm. By using edge information expression structure encoding unit complexity, using edge information difference between the current encoding unit and the spatial adjacent encoding unit to reflect encoding units within the edge similarity and edge mobility, and the distortion generation value of combining depth information and prediction unit edge similarity and edge mobility and spatial adjacent encoding unit rate, advance judge encoding unit segmentation form and prediction unit model. The experimental results show that this algorithm is a fast inter frame encoding, In the random access mode (RA) and low latency B frames (LD-B) mode, in comparison with the original HEVC encoding algorithm, an average of 53.7% and 54.9% respectively improve the encoding speed of.2. and presents a fast encoding algorithm of HEVC based on video content, video content analysis, it can reflect the 3 kinds of video content parameters are the mean pixel encoding unit, and the variance of the edge gradient magnitude, and encoding information with adjacent encoding unit combined with fast encoding unit size and fast encoding mode selection unit. The experimental results show that the proposed fast algorithm in RA mode and LD-B mode, a HEVC encoding algorithm. An average of 58.9% and 57.6% increase the encoding speed of.3. presents a fast encoding algorithm of 3D-HEVC based on the depth chart, to reduce the texture video encoding complexity. We will be in this depth map fast algorithm The mean square error of different depth value encoding unit in the number and depth value as two parameters, and through these two parameters to achieve early to judge the cell size is decided in advance and the encoding mode of Skip. Finally, the research results of this paper, and forecasts the video encoding field directions for future research.

【学位授予单位】：吉林大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TN919.81

【相似文献】