VP9帧内预测快速算法研究

发布时间：2018-04-26 11:53

本文选题：VP9 + 视频编码　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：VP9是谷歌继推行VP8后于2013年推行的新一代开源视频编码标准。VP9视频编码标准保留了原有的传统的视频编码框架,同时采纳了一系列新技术,使其在同等视频质量下码率较上一代VP8可降低50%。但是压缩性能提高的同时也带来了巨大的编码复杂度。本文针对这一问题研究了VP9视频编码标准的帧内预测快速算法,以期望降低编码复杂度提高编码速度,使该标准更适合实时性要求较高的业务。为了达到降低编码复杂度的目的,本文首先对VP9视频编码的帧内预测过程进行了深入的研究和分析;然后在此基础上,提出了两种适用于VP9的帧内预测快速算法。分别是帧内预测模式快速选择算法和帧内预测的块划分快速算法。本文针对VP9帧内预测中模式选择复杂度高的问题,提出了一种帧内预测模式快速选择算法。每一个64×64的超级块需要进行6810次模式选择才能确定最佳划分方式以及相应的最佳预测模式。通过减少预测模式的候选模式数目可以大大减少模式选择的次数。本文通过利用相邻预测块的最佳预测模式和当前块的纹理方向性来减少帧内预测的候选模式。该算法排除了大部分预测模式,从而降低了编码复杂度,节约了近55%的编码时间。针对VP9帧内预测中块划分过程较复杂的问题,提出了一种帧内预测的块划分快速算法。块划分过程算法优化的大体思路为直接确定某些块的最优划分方式或者跳过某些不必要的块划分方式。首先,确定了影响最优块划分的因素:空间复杂度、空间平滑度和编码量化参数。然后,根据这3个因素和帧内预测中块划分的关系,对平坦的预测块只进行较大块预测,而对不平坦的预测块则直接跳过大块预测只进行较小块预测。实验证明该算法对于运动和纹理复杂度不同的序列,在不同的量化参数下,都能够在保证视频编码效率下降较小的同时,大幅度降低帧内编码的复杂度从而使编码时间得到大幅度减少。
[Abstract]:VP9, Google's new open-source video coding standard. VP9, introduced in 2013 after VP8, retains the traditional video coding framework and incorporates a range of new technologies. Compared with the previous generation VP8, the bitrate can be reduced by 50% under the same video quality. However, the compression performance is improved, but also brings huge coding complexity. In order to reduce the coding complexity and improve the coding speed, the fast intra prediction algorithm of VP9 video coding standard is studied in this paper. In order to reduce the coding complexity, this paper firstly studies and analyzes the intra prediction process of VP9 video coding, and then proposes two fast intra prediction algorithms for VP9. It is a fast selection algorithm for intra prediction mode and a fast block partition algorithm for intra prediction. Aiming at the high complexity of mode selection in VP9 intra prediction, this paper proposes a fast mode selection algorithm for intra prediction. Each 64 脳 64 superblock needs 6810 mode selections to determine the optimal partitioning method and the corresponding optimal prediction model. By reducing the number of candidate patterns, the number of pattern selections can be greatly reduced. In this paper, the best prediction mode of adjacent prediction block and the texture directivity of current block are used to reduce the candidate mode of intra prediction. The algorithm eliminates most prediction modes, thus reducing the coding complexity and saving about 55% coding time. A fast block partitioning algorithm for intra prediction of VP9 is proposed to solve the complex problem of block partitioning in intra prediction. The general idea of block partition algorithm optimization is to determine the optimal partition method of some blocks directly or skip some unnecessary block partition methods. Firstly, the factors affecting the optimal block partition are determined: space complexity, spatial smoothness and coding quantization parameters. Then, according to the relationship between these three factors and block partition in intra prediction, only large blocks are predicted for flat prediction blocks, while smaller blocks are predicted for uneven prediction blocks directly skipping over large blocks. Experiments show that the proposed algorithm can reduce the video coding efficiency under different quantization parameters for sequences with different motion and texture complexity. The complexity of intra-coding is greatly reduced and the coding time is greatly reduced.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN919.81

【参考文献】