基于复杂度差异的VVC中360度视频CU划分快速决策  

Fast CU Partition Decision Based on Complexity Differences for 360 Degree Video with VVC

作　　者：张晨昕 ZHANG Chenxin(North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《北京工业职业技术学院学报》2021年第3期32-38,共7页Journal of Beijing Polytechnic College

摘　　要：作为新一代视频编码标准,H.266/VVC引入了嵌套在多类型树中的四叉树划分结构,但编码时间明显增加。在360度视频场景下,针对上述问题,设计一种CU划分快速决策算法。通过统计分析不同纬度CU划分结构的分布特征,发现不论多叉树如何进行划分,1个CU的四叉树深度可以代表它的复杂程度。在各个纬度区分简单与复杂CU,分别决定两者可达的四叉树深度,由四叉树深度抉择四叉树划分与多叉树划分的选取。另一方面通过CU子块复杂度的差异辅助判断复杂CU的划分方式,可以降低原有算法的计算量,以节省编码时间。实验结果表明:在全I帧模式下,与现有的VTM8.0相比,时间节省了24%,WS-PSNR降低了0.02 dB,BD-rate仅增加了0.49%。As a new generation of video coding standard,H.266/VVC introduces a partition structure of quadtrees nested in Multi-Type trees,but the coding time increases significantly.In the 360 degree video scene,aiming at the above problems,this paper designs a fast algorithm for CU partition.By statistically analyzing the distribution characteristics of CU partition structure in different latitudes,this paper finds that regardless of the partitioning of binary tree and ternary tree,the quadtree depth of a CU can represent its complexity.In each latitude,the simple and complex CU can be distinguished,and the depth of quadtree can be determined respectively.The choice of quadtree Partition and multi tree partition is determined by the depth of quadtree.On the other hand,the complexity difference of CU sub blocks is used to determine the division mode of complex CU,which reduces the calculation of the original algorithm and saves the coding time.The experimental results show that:compared with the existing VTM 8.0,the time is saved by 24%,the WS-PSNR is reduced by 0.02 dB,and the BD-rate is only increased by 0.49%.

关 键 词：VVC 360度视频 多叉树 纹理复杂度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]