基于视频内容复杂度的视频编解码算法研究

发布时间：2018-06-28 08:04

本文选题：高效视频编码 + 图像复杂度　；参考：《北方工业大学》2017年硕士论文

【摘要】：新一代视频编码标准——高效视频编码标准(HEVC,High efficiency video coding)的提出,显著提升了编码效率,受到研究人员的广泛重视。HEVC编码标准中,引入了新的四叉树编码结构,在提高编码效率的同时也在一定程度上的增加了编码的复杂度,影响了编码器的实时性。此外,帧内块拷贝算法的引入,提升了 HEVC屏幕内容编码的效率,但是该工具采用了固定尺寸的匹配块搜索方法,降低了匹配块的搜索能力。针对以上问题,论文在分析视频内容复杂度的基础上,从降低四叉树编码复杂度,提高帧内块拷贝算法匹配效率两个方面开展研究,提升HEVC编码器的编码性能。论文的主要创新点包括以下两个方面。第一,针对现有HEVC框架的四叉树结构编码复杂度高的问题,深入研究了编码单元四叉树递归划分过程,发现了编码单元大小与该单元图像复杂度之间的相关性,提出了利用图像复杂度来跳过率失真优化过程或提前判定编码单元划分的低复杂度算法,算法在基本不影响视频压缩质量的情况下明显降低了编码的复杂度。第二,研究了屏幕内容的帧内块拷贝技术,发现该工具由于采用固定尺寸的匹配块搜索方法,对一些复杂连续文字区域的匹配块搜索能力较弱,降低了编码效率。论文设计了一种基于自适应尺寸的匹配块搜索方法,以方差的变化为指标对编码单元进行自适应块划分,使得块的尺寸和块内容的复杂度成反比,从而有利于提高匹配度和编码效率。针对以上所提两个算法,进行了大量的实验进行测试,实验结果表明,论文所提第一种算法,在All Intra(AI)编码条件下,降低了平均35%的编码时间;所提的第二种算法,在All intra(AI)编码条件下带来了0.96%视频重建质量的提升的同时只增加了平均3%的编码时间。
[Abstract]:The new generation video coding standard, HEVC High efficiency video coding) (High Video coding Standard), has greatly improved the coding efficiency. It has been widely paid attention to by researchers, and a new quadtree coding structure has been introduced. At the same time, it also increases the complexity of encoder to some extent and affects the real-time performance of encoder. In addition, the introduction of intra-frame block copy algorithm improves the efficiency of HEVC screen content coding, but the tool uses fixed-size matching block search method, which reduces the search ability of matching block. In order to improve the coding performance of HEVC encoder, this paper analyzes the complexity of video content and improves the coding performance of HEVC encoder by reducing the complexity of quadtree coding and improving the matching efficiency of intra-frame block copy algorithm. The main innovation of the paper includes the following two aspects. First, aiming at the high complexity of the existing HEVC frame, the paper deeply studies the quadtree recursive partition process of the coding unit, and finds the correlation between the size of the coding unit and the complexity of the unit image. A low complexity algorithm using image complexity to skip the rate-distortion optimization process or determine the partition of coding units in advance is proposed. The algorithm reduces the coding complexity significantly without affecting the quality of video compression. Secondly, the intra-frame block copy technology of screen content is studied. It is found that the matching block search method of fixed size is used in this tool, and the matching block search ability of some complex continuous text regions is weak, which reduces the coding efficiency. In this paper, an adaptive block search method based on adaptive size is designed. The adaptive block partition is based on variance variation, which makes the block size inversely proportional to the complexity of block content. Therefore, it is helpful to improve the matching degree and coding efficiency. A large number of experiments have been carried out to test the two algorithms mentioned above. The experimental results show that the first algorithm can reduce the average coding time by 35% under the condition of All intra (AI) coding. Under the condition of All intra (AI, the quality of video reconstruction is improved by 0.96% and the average coding time is only increased by 3%.
【学位授予单位】：北方工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.81

【参考文献】