SHVC质量可伸缩视频编码的快速算法研究

项目名称： SHVC质量可伸缩视频编码的快速算法研究

项目编号： No.61401247

项目类型： 青年科学基金项目

立项/批准年度： 2015

项目学科： 电子学与信息系统

项目作者： 汪大勇

作者单位： 清华大学

项目金额： 8万元

中文摘要： 随着网络技术和终端处理能力的不断提高,高清视频是当前发展的趋势, 为了适应网络带宽的变化和终端的多样性，基于HEVC 的可伸缩视频编码标准SHVC 成为了研究的热点。SHVC 具有很高的编码效率，但其编码复杂度也非常高。研究基于SHVC 的快速编码算法具有重要的理论意义与广泛的应用前景。本项目旨在研究SHVC 质量可伸缩视频编码的快速算法：(1)研究方向模式与对应哈达玛值之间的关系及起始方向模式的选择，在此基础上得到最优方向模式；(2)研究层间预测单元分布规律，编码可能性很大的预测单元并对得到的残差系数研究是否满足高斯分布，从而得到预测单元提前终止判断条件；(3)研究层间编码深度分布规律，编码可能性较大的深度，跳过可能性很小的深度，根据残差或者纹理决定是否编码可能性较小的深度。通过以上研究，在编码效率损失很小的情况下编码速度能提高70%以上，从而推动我国信息产业的进一步发展。

中文关键词： SVC；相关性；模式预测；提前终止；

英文摘要： With the development in network technology and the terminal processing capability, high-definition video becomes the direction of developmenttrend. In order to adapt to variations in network bandwidth and the diversity of terminals, scalable video coding based on HEVC standard, so called SHVC (scalable HEVC extension) has become a hot research field.SHVC targets high coding efficiency,but its coding complexity is also very high, which seriously affects its future application. So, investigating fast encoding algorithms for SHVC has high research significance and broad application prospect. This proposal plans to study on fast algorithms for SHVC quality scalable video coding:(1)study on the relationship between direction modes and Hadamard Cost,the selection of initial direction modes,and then the best direction modes are obtained on this basis;(2)study on the distribution of inter-layer prediction units,decide whether the residual coefficients of pedicetion units with high possibilities is obeyed Gauss distrubition,so as to obtaind the early termination of prediction units;(3)study on the distribution of coding depths,encode the depth with high possibilities,skip the depth with very low possibilities and decide whether to encode the depth with relative low possibilities according to the differences of residuals

英文关键词： SVC；correlation；mode prediction；early termination；