面向自由视点视频系统的深度视频传输方法  

Depth Video Transmission Scheme for Free Viewpoint Video System

作　　者：余思文[1] 彭宗举[1] 郁梅[1] 蒋刚毅[1] 邵枫[1] 陈芬[1] 

机构地区：[1]宁波大学信息科学与工程学院,宁波315211

出　　处：《计算机辅助设计与图形学学报》2013年第12期1826-1833,共8页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61171163;61271270);国家自然科学基金国际合作与交流项目(61111140392);宁波市自然科学基金(2012A610041);新一代移动互联网用户端软件科技创新团队(2010R50009)

摘　　要：针对自由视点视频系统中深度视频数据量大、深度编码失真影响虚拟视点绘制图像质量的问题，提出一种深度视频传输方法．该方法包括深度视频预处理和后处理两部分：在深度视频预处理阶段建立绘制无失真模型，并基于该模型对深度图进行高斯平滑及下采样以减少编码比特数；在后处理阶段，对解码和上采样后的深度视频采用自适应边缘重建以恢复深度边缘信息．实验结果表明，文中方法能节省53．87％～73．62％的码率，提高了虚拟视点绘制质量．A depth video transmission scheme is proposed to solve the problems in free viewpoint video system, huge data amount of depth video and effect of depth video coding distortion on virtual view rendering. This scheme contains two parts, pre-processing and post-processing. During the pre- processing part, depth no-synthesis-error model is established and utilized to aid Gaussian filter. In addition, depth video is down-sampled to reduce the bit rate. During the post-processing part, the decoded and up-sampled depth video is processed by adaptive depth edge reconstruction filter. Experimental results show that the proposed method can save bit rate ranging from 53. 87% to 73.62%, while it improves the overall rendering quality.

关 键 词：深度视频 虚拟视点绘制 绘制无失真 深度边缘重建 

分 类 号：TN919.81[电子电信—通信与信息系统]