| 講演名 | 2006/1/2 CLASSIFICATION OF REACHING ACTION FOR MOTION PARAMETERIZATION(International Workshop on Advanced Image Technology 2006) , |
|---|---|
| PDFダウンロードページ | PDFダウンロードページへ |
| 抄録(和) | |
| 抄録(英) | In order to generate controllable human animation, several methods based on the interpolation of motion capture data have been proposed. These motion parameterization methods require a data set which consists of similar motion to avoid artifacts in generated motion. However, reaching for instance, has considerable variation even for the same reaching position including the difference of the direction of hand approach and the foot to step forward. Therefore, to generate natural motion under unrestricted conditions, an automatic classifying technique is needed before interpolating motions. In our method, we adopt the Laplacian eigenmaps technique for classifying reaching motions. First, Laplacian eigenmaps method is applied to reduce the dimension of the postures sampled from the motion data, preserving local structure of the original high dimensional data. Then, k-means algorithm is used in the low dimension space to cluster the samples. At last, reaching motions are classified according to the clustering results. We applied our method to reaching data parameterized by the orientation of a tube shaped target object, and synthesized motion from classified motions successfully. Our technique enables us to apply the motion parameterization techniques to more general condition which results in more natural animation. |
| キーワード(和) | |
| キーワード(英) | |
| 資料番号 | IE2005-141 |
| 発行日 |
| 研究会情報 | |
| 研究会 | IE |
|---|---|
| 開催期間 | 2006/1/2(から1日開催) |
| 開催地（和） | |
| 開催地（英） | |
| テーマ（和） | |
| テーマ（英） | |
| 委員長氏名（和） | |
| 委員長氏名（英） | |
| 副委員長氏名（和） | |
| 副委員長氏名（英） | |
| 幹事氏名（和） | |
| 幹事氏名（英） | |
| 幹事補佐氏名（和） | |
| 幹事補佐氏名（英） | |
| 講演論文情報詳細 | |
| 申込み研究会 | Image Engineering (IE) |
|---|---|
| 本文の言語 | ENG |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | CLASSIFICATION OF REACHING ACTION FOR MOTION PARAMETERIZATION(International Workshop on Advanced Image Technology 2006) |
| サブタイトル（和） | |
| キーワード(1)（和/英） | |
| 第 1 著者 氏名（和/英） | / Kouji SATOU |
| 第 1 著者 所属（和/英） | Department of Computer Science, School of Engineering |
| 発表年月日 | 2006/1/2 |
| 資料番号 | IE2005-141 |
| 巻番号（vol） | vol.105 |
| 号番号（no） | 500 |
| ページ範囲 | pp.- |
| ページ数 | 5 |
| 発行日 | |