| 大会名称 |
|---|
| 2009年 情報科学技術フォーラム(FIT) |
| 大会コ－ド |
| F |
| 開催年 |
| 2009 |
| 発行日 |
| 2009/8/20 |
| セッション番号 |
| 3J |
| セッション名 |
| 動画像解析 |
| 講演日 |
| 2009/09/02 |
| 講演場所(会議室等) |
| J会場（9号館2F 924教室） |
| 講演番号 |
| H-039 |
| タイトル |
| Motion Beat Induction by Short-Term Principal Component Analysis |
| 著者名 |
| Xu Jianfeng, Takagi Koichi, Yoneyama Akio, |
| キーワード |
| human motion, beat induction, Short-Term Principal Component Analysis, dimension reduction |
| 抄録 |
| A novel short-term principal component analysis (ST-PCA) method is proposed to estimate motion beats in human motion, which is a high dimensional time series. ST-PCA performs PCA in a sliding window to locally extract the major variance of the movement into the coordinates of the first principal component, revealing the desired motion beats accurately. Our algorithm is carefully designed for the three characteristics in human motion: hierarchical structure, spatial correlation, and temporal coherence. In our experiments, the proposed method obtains much more accurate motion beats in a wide range of motion categories including complicated dances than the state-of-the-art. |
| 本文pdf |
| PDF download (388.4KB) |