| Project/Area Number |
17500067
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Media informatics/Database
|
|---|
| Research Institution | Shizuoka University |
|---|
Principal Investigator |
KANEKO Toru Shizuoka University, Faculty of Engineering, Professor, 工学部, 教授 (50293600)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YAMASHITA Atsushi Shizuoka University, Faculty of Engineering, Assistant Professor, 工学部, 助手 (30334957)
ARAI Tamio The University of Tokyo, Faculty of Engineering, Professor, 工学部, 教授 (40111463)
MIURA Kenjiro Shizuoka University, Graduate School of Science and Technology, Professor, 創造科学技術大学院, 教授 (50254066)
OKUDAIRA Masashi Musashi InstituteofTechnology, Faculty of Environmental and Information Studies, Professor, 環境情報学部, 教授 (00350231)
|
|---|
| Project Period (FY) |
2005 – 2006
|
| Project Status |
Completed (Fiscal Year 2006)
|
| Budget Amount *help |
¥3,600,000 (Direct Cost: ¥3,600,000)
Fiscal Year 2006: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 2005: ¥2,600,000 (Direct Cost: ¥2,600,000)
|
|---|
| Keywords | Image processing / Image composition / Chroma key / Stereo image / Computer vision / 3-dimensional measuremtn / Virtual reality / Stereo camera |
|---|
| Research Abstract |
1. Object extraction from images captured with a stripe background: A method for object extraction was constructed to create a composite of two images, one is that of objects taken at a studio, and the other is that of a background scene taken at another place and time. The method uses a two-tone stripe screen to solve the problem that conventional chroma key cannot extract an object whose color is the same as the background screen
2. Improvement of contour extraction : A technique of active contours (Snakes) is employed to make the extraction results given by stripe chroma key more precise.
3. Object extraction from images captured with a checker pattern background: A stripe background has a directionality which makes it impossible to extract contours in the same direction as the stripes. In order to solve this problem, we proposed a method using a checker pattern background which has both horizontal and vertical edges..
4. Improvement of image composition: A method for determining the so-called alpha value was constructed because a composite image quality depends on the alpha value which is a measure of the ratio of mixture of the foreground and background pixel values on the boundaries of foreground object contours.
5. Calculation of geometrical relation for image composition: Ignorance of relations of positions and scales between foreground objects and backgrounds gives unnatural composite images. Therefore, a method for obtaining geometrical information of foreground objects was investigated by employing 3-dimensional measurement system based on stereo vision.
Experiment of stereo image composition: The validity of the proposed method for object extraction and compositing was confirmed by experiment with many foreground images.
|
