Learning to Reconstruct 3D Structure from Object Motion | |
Liu, Wentao ; Dou, Haobin ; Wu, Xihong | |
2015 | |
关键词 | 3D Reconstruction Structure from Motion Deep Neural Network Kinetic Depth Effect IMAGE STREAMS FACTORIZATION ALGORITHM FEATURES SHAPE |
英文摘要 | In this paper, we propose a new approach for reconstructing 3D structure from motion parallax. Instead of obtaining 3D structure from multi-view geometry or factorization, a Deep Neural Network (DNN) based method is proposed without assuming the camera model explicitly. In the proposed method, the targets are first split into connected 3D corners, and then the DNN regressor is trained to estimate the relative 3D structure of each corner from the target rotation. Finally, a temporal integration is performed to further improve the reconstruction accuracy. The effectiveness of the method is proved by a typical experiment of the Kinetic Depth Effect (KDE) in human visual system, in which the DNN regressor reconstructs the structure of a rotating 3D bent wire. The proposed method is also applied to reconstruct another two real targets. Experimental results on both synthetic and real images show that the proposed method is accurate and effective.; EI; CPCI-S(ISTP); liuwt@cis.pku.edu.cn; douhb@cis.pku.edu.cn; wxh@cis.pku.edu.cn; 127-137; 9489 |
语种 | 英语 |
出处 | NEURAL INFORMATION PROCESSING, PT I |
DOI标识 | 10.1007/978-3-319-26532-2_15 |
内容类型 | 其他 |
源URL | [http://ir.pku.edu.cn/handle/20.500.11897/436922] |
专题 | 信息科学技术学院 |
推荐引用方式 GB/T 7714 | Liu, Wentao,Dou, Haobin,Wu, Xihong. Learning to Reconstruct 3D Structure from Object Motion. 2015-01-01. |
个性服务 |
查看访问统计 |
相关权益政策 |
暂无数据 |
收藏/分享 |
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。
修改评论