基于最大可容忍深度失真模型的低复杂度深度视频编码  

Low complexity depth video coding based on maximum tolerable depth distortion model

作　　者：李祖团 邵枫[1] 蒋刚毅[1] 郁梅[1] 李福翠[1] 

机构地区：[1]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》2015年第6期1162-1169,共8页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61271021)资助项目

摘　　要：在保证虚拟视点绘制质量的前提下,如何降低深度视频的编码复杂度是一个亟需解决的问题。本文提出了一种基于最大可容忍深度失真(MTDD,maximum tolerable depth distortion)模型的低复杂度深度编码算法,将MTDD模型引入到率失真(RD)代价函数,通过对编码单元(CU)进行失真度量判决,如果对当前CU进行编码不会导致可察觉的绘制失真,则失真度量只考虑深度编码失真,否则失真度量同时考虑深度编码失真和视点合成失真(VSD),从而降低编码复杂度。实验结果表明,所提出的算法在不降低虚拟视点绘制质量和不增加编码码率的情况下,能显著降低深度视频的编码复杂度。How to reduce the computational complexity for depth video coding while maintaining high rein dering quality of synthesized virtual views is still an urgent problem to be solved. In this paper, a low complexity depth video coding method is proposed based on maximum tolerable depth distortion （MTDD） model. Specifically,the MTDD model is incorporated into the rate-distortion （RD） function. To determine distortion measurement for each coding unit （CU） ,if the depth coding distortion will not lead to noticeable view synthesis distortion,only depth coding distortion term is considered in distortion measurement, otherwise, both depth coding distortion and view synthesis distortion are considered. Experimental results show that the proposed algorithm can significantly reduce the coding complexity while without obvious rendering quality degradation or coding bitrate increasement.

关 键 词：3D视频 深度视频编码 率失真(RD)代价 最大可容忍深度失真(MTDD) 

分 类 号：TN919.8[电子电信—通信与信息系统]