3次元ビデオの時間的空間的編集
Spatial and Temporal Motion Editing for Interaction Scenes in 3D Video
概要
3D video is a real movie recording dynamic visual events in the real world as is: time varying 3D object shape with high fidelity surface texture. In capturing multi-object interaction scenes, multiple actors are usually required to be captured separately to avoid occlusions and some other problem, which will inevitably result in spatial and temporal mismatches. We proposed a method to realize spatiotemporally synchronized 3D video synthesis from independently captured motion sequences. Our main idea is to introduce Motion History Volume (MHV) for multi-object interaction representation and categorization. By doing constraint satisfaction based spatiotemporal alignment and path optimization, we can generate natural multi-object interaction scenes from separately captured data.
産業界への展開例・適用分野
The potential applications of our work cover wide varieties of fields. For entertainment, it can be used in 3D game industry or movie industry for virtual character editing; As for education and business, it can be applied to teleconference system to help people communicate better; In addition, it can play an important role in the work of preserving cultures. (e.g. 3D archive of traditional dances or martial arts).
研究者
| 氏名 | 専攻 | 研究室 | 役職/学年 |
|---|---|---|---|
| 石 群 | 知能情報学専攻 | 松山研究室 | 博士2回生 |
| 延原 章平 | 知能情報学専攻 | 松山研究室 | 講師 |
| 松山 隆司 | 知能情報学専攻 | 松山研究室 | 教授 |