2017 Fiscal Year Final Research Report
Development of a support system for learning martial arts techniques by video observation based on 3D point cloud processing
| Project/Area Number |
15K01102
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Educational technology
|
|---|
| Research Institution | Kindai University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Renkei-kenkyūsha) |
OCHI Youji 近畿大学, 理工学部, 准教授 (80314847)
|
|---|
| Project Period (FY) |
2015-04-01 – 2018-03-31
|
| Keywords | スポーツ / 振り返り学習 / ビデオ映像 / シーン検出 / 3D点群 / ポイントマッチング / ホモグラフィ変換 / 三次元復元 |
|---|
| Outline of Final Research Achievements |
We have developed a pose detection method by three dimensional (3D) point cloud matching between a player and a sample, and also have developed an offensive motion detection method based on the time rate of change of the variance of point cloud of two players. Thus, the study makes it possible to detect objective scene from video images taken with RGBD-camera for learning martial arts techniques by video observation.
Next, in order to enable motion observation from an arbitrary angle, we developed a basic method for generating a 3D model of a player by estimating the three dimensional coordinates of the player's joints from video images. We also developed a method to generate bird's-eye view images of foot positions on a floor by the homography transformation for the observation of players' footwork.
|
| Free Research Field |
教育工学
|
