北京航空航天大学学报 ›› 2014, Vol. 40 ›› Issue (6): 749-755.doi: 10.13700/j.bh.1001-5965.2013.0437

• 论文 • 上一篇    下一篇

基于分段化结构重建的2D/3D视频转换方法

刘伟1, 李功燕1, 丁锴2, 吴毅红3   

  1. 1. 中国科学院微电子研究所 物联网中心, 北京 100029;
    2. 北京航空航天大学 自动化科学与电气工程学院, 北京 100191;
    3. 中国科学院自动化研究所 模式识别国家重点实验室, 北京 100190
  • 收稿日期:2013-07-30 出版日期:2014-06-20 发布日期:2014-07-03
  • 作者简介:刘 伟(1983- ),男,河南南阳人,助理研究员,lw3171796@163.com.

2D/3D video conversion method based on piece-wise structure from motion

Liu Wei1, Li Gongyan1, Ding Kai2, Wu Yihong3   

  1. 1. Center for Internet of Things, Institute of Microelectronics of Chinese Academy of Sciences, Beijing 100029, China;
    2. School of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100191, China;
    3. National Laboratory of Pattern Recognition, Institute of Automation of Chinese Academy of Sciences, Beijing 100190, China
  • Received:2013-07-30 Online:2014-06-20 Published:2014-07-03

摘要: 为了缓解3D片源的不足,设计了一种分段化结构重建方法为视频2D/3D转换提供深度线索.采用分层提取的方式,将视频基于内容分解为子序列,在每个子序列中建立优化准则选择出处于非退化状态的关键帧,并以此为基础提出了分段化结构重建框架,利用改进的耦合自标定算法重构出每个子序列对应的离散3D结构信息,从而获取视频的深度线索.实验结果表明,该方法通过分段化后的局部优化,不仅使结构化重建能够有效处理多场景下的连续镜头,还加快了转换的效率,能够稳定、高效地获取深度信息,适用于视频2D/3D转换.

Abstract: To alleviate the shortage of 3D media sources, a piece-wise structure from motion method was designed to provide depth cues for the 2D/3D video conversion. Videos were decomposed into subsequences of different scenes and key frames of non-degenerate were extracted in each subsequence with optimization criteria. On this basis, a piece-wise structure from motion framework was proposed and depth cues of scene sparse structures were obtained through an improved coupling self-calibration algorithm. Experimental results show that by local optimization after video segmentation, the proposed method can handle video shot containing multiple scenes based on structure from motion, and improve the efficiency of conversion effectively. Those mean the proposed method is suitable for the 2D/3D video conversion.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发