2003 Fiscal Year Final Research Report Summary
Three-dimensional Modeling of Dynamic Environments Employing Mobile Stereo Vision
| Project/Area Number |
14580450
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報システム学(含情報図書館学)
|
|---|
| Research Institution | Kyushu Institute Of Technology |
|---|
Principal Investigator |
ISHIKAWA Seiji Kyushu Institute Of Technology, Department of Control Engineering, Professor, 工学部, 教授 (90128116)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TAN Joo kooi Kyushu Institute Of Technology, Department of Control Engineering, Research Associate, 工学部, 助手 (40363395)
|
|---|
| Project Period (FY) |
2002 – 2003
|
| Keywords | Motion capturing / Camera calibration / Three-dimensional recovery / Stereoscopic cameras / Stereoscopic vision / Factorization |
|---|
| Research Abstract |
Three-dimensional modeling of human motions is important, since it has many application fields such as rehabilitation, sports, video game production, etc. In order to perform three-dimensional modeling, the motion concerned normally needs to be acted in a large studio equipped with a number of settled cameras This excludes various motions (motions observed outdoors, for example) from being modeled, however. There hasn't been an established modeling. technique that can model those motions which are acted in various places including outdoor environments.
In this situation, the present research aims at achieving the followings :
(1)To develop a three-dimensional shape/motion recovery technique based on more than two mobile cameras that capture images while moving.
(2)To develop a technique for three-dimensional modeling of a remote dynamic environment where people have various actions by sending the images using an image transfer apparatus.
We have developed a technique for modeling a three-dimensional object based on image capture by mobile cameras. In the first place, we habe established a method of camera calibration using two-dimensional static feature points extracted on the obtained video images. Then we have developed a method of computing three-dimensional positions of the points on an object concerned in order to model shape/motion of the object.
In the neat step, employing the above technique, we have developed a system for modeling a remote dynamic environment whose images are captured by mobile cameras and sent by image transfer apparatus. Performed experiments gave satisfactory results.
|
