基于HEVC帧内预测算法的优化

发布时间：2018-06-27 05:48

本文选题：HEVC + 帧内预测　；参考：《南京理工大学》2014年硕士论文

【摘要】：高效率视频编码标准(HEVC,High Efficiency Video Coding)于2010年由JCT-VC制定,并于2013年正式发布。相比较上一代视频压缩标准H.264/AVC,在相同图像质量下HEVC能节省50%的码率,但是计算复杂度却大幅提高了。因此如何降低HEVC视频编码算法的计算复杂度,提高编码的时效性成为了当前研究的热点。鉴于此,本文重点研究了帧内编码算法及其优化方法,主要工作及成果如下： 1)针对最大编码单元(LCU, Largest Coding Unit)中,编码块划分次数过多导致的高计算复杂度问题,本文提出了一种提前终止编码块划分的优化算法。该算法首先利用图像的纹理特性,分析了不同深度下相邻编码块划分的相关性；然后提出了提前终止编码块划分的方案：即利用左侧、上方、左上方已编码块的划分信息,来决定右下方编码块是否需要划分。实验表明：本文提出的算法与原始算法相比较,码率增加了约1.02%,峰值信噪比下降了约0.045dB,编码时间缩短了近17.79%。 2)本文在研究了帧内预测中模式选择算法的基础上,提出了一种快速预测模式选择的优化算法。该算法将粗略模式选择(RMD, Rough Mode Decision)过程,分成初筛选和再筛选两步,以此减少了RMD模式的个数；然后利用最有可能模式(MPM, Most Probable Mode)和第一个候选模式的相关性,提出了优先选取最佳预测模式的方案；最后对候选模式进行分组,自适应地减少了参与率失真代价计算的模式个数。实验表明：本文提出的算法与原始算法相比较,码率增加了约0.04%,峰值信噪比下降了约0.034dB,编码时间缩短了近25.1%。最后本文将以上两种算法合并成了一个新的优化算法。新算法相比较HEVC原始算法,码率增加了约1.08%,峰值信噪比下降了约0.075dB,编码时间减少了近37.16%。
[Abstract]:The HEV C High efficiency Video coding standard was developed by JCT-VC in 2010 and released in 2013. Compared with the previous video compression standard H.264 / AVC, HEVC can save 50% bit rate under the same image quality, but the computational complexity is greatly increased. Therefore, how to reduce the computational complexity of HEVC video coding algorithm and improve the timeliness of coding has become a hot topic. In view of this, this paper focuses on the intra coding algorithm and its optimization method. The main work and results are as follows: 1) in order to solve the problem of high computational complexity caused by the excessive number of coding blocks in LCU (Largest coding Unit), In this paper, an optimization algorithm for preemptive termination of block partition is proposed. The algorithm firstly analyzes the correlation between adjacent coding blocks at different depths by using the texture characteristics of the image, and then puts forward a scheme to terminate the partition of coding blocks in advance, that is, using the partition information of the coded blocks on the left, the top and the upper left. To determine whether the lower right coding block needs to be partitioned. Experimental results show that compared with the original algorithm, the proposed algorithm increases the bit rate by about 1.02db, decreases the PSNR by 0.045dB, and shortens the coding time by nearly 17.79dB. 2) based on the study of the mode selection algorithm in intra prediction, An optimization algorithm for fast prediction mode selection is proposed. The algorithm reduces the number of RMD modes by dividing the rough Mode decision process into two steps: initial filtering and re-filtering, and then takes advantage of the correlation between MPM (most enable Mode) and the first candidate mode. The optimal prediction mode is selected first and the candidate modes are grouped to reduce the number of modes of participating rate-distortion cost calculation adaptively. Experimental results show that compared with the original algorithm, the proposed algorithm increases the bit rate by about 0.04, decreases the PSNR by about 0.034dB, and shortens the coding time by about 25.1dB. Finally, the above two algorithms are combined into a new optimization algorithm. Compared with the original HEVC algorithm, the new algorithm increases the bit rate by about 1.08, decreases the peak signal-to-noise ratio (PSNR) by 0.075 dB, and reduces the coding time by about 37.16dB.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN919.81

【参考文献】