基于转移概率的多视点视频快速模式选择算法  

Fast mode decision for multiview video coding based on transition probability

在线阅读下载全文

作  者:李跃[1] 朱亚培[2] 蒋刚毅[3] 刘灿[2] 游开明[2] 

机构地区:[1]湖南工学院,湖南衡阳421002 [2]衡阳师范学院物理与电子信息科学系,湖南衡阳421002 [3]宁波大学信息科学与工程学院,浙江宁波315211

出  处:《光电子.激光》2014年第12期2379-2386,共8页Journal of Optoelectronics·Laser

基  金:湖南省自然科学基金(12JJ6001)资助项目

摘  要:针对多视点视频编码(MVC)中部分大宏块模式(SKIP/DIRECT模式,帧间16×16模式)相对于其它模式的计算复杂度低,并且在编码所确定的最优模式中占有相当高比重的特点,本文提出一种基于转移概率的多视点视频快速模式选择算法。首先根据多视点视频当前编码宏块对应时间和视点间宏块及周围宏块的位置几何关系,建立宏块模式参考模型;其次比较参考模型中模式的时间相似度和视点间相似度;最后计算模式之间的转移概率,确定提前结束大宏块模式选择过程的阈值。实验结果表明,本文所提出的快速算法平均节约MVC时间80.93%,编码质量平均下降0.04dB,码率平均增加0.27%。The multiview video coding (MVC) adopts the exhaustive mode decision to achieve high effi- ciency,at the price of extremely high computational complexity. Compared with the macroblock (MB) mode,the large MB mode (SKIP/DIRECT, inter 16 × 16) occupies most of the MB mode proportion, while its estimation consumes much less computational complexity than that of other modes. By using temporal and interview correlation between current MB and its adjacent MBs,a fast mode decision for MVC based on transition probability is proposed'in this paper. By utilizing the geometric relationship between the current MB and the corresponding coded MBs in the temporal and the interview frames, the MB mode reference model is firstly established. Compared with the temporal and interview mode similarity of the mode reference model, the transition probability between two modes is secondly calculated. Lastly, the threshold is obtained to terminate the process of the large MB mode decision. The experimen- tal results show that compared with the exhaustive mode decision in MVC,the fast mode decision algorithm for MVC can achieve a reduction of the computational complexity by 80.93% in average encoding time,while incurring only 0.04 dig loss in peak signal-to-noise ratio (PSNR) and 0. 27% increment in the total bit rate (BR).

关 键 词:多视点视频编码(MVC) 模式选择 转移概率 

分 类 号:TN919.81[电子电信—通信与信息系统]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象