| Project/Area Number |
13680431
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Utsunomiya University |
|---|
Principal Investigator |
SHOJI Kenji Utsunomiya University, Faculty of Engineering, Associate Professor, 工学部, 助教授 (70143188)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TOYAMA Fubito Utsunomiya University, Faculty of Engineering, Assistant Professor, 工学部, 助手 (60323317)
|
|---|
| Project Period (FY) |
2001 – 2002
|
| Project Status |
Completed (Fiscal Year 2002)
|
| Budget Amount *help |
¥4,200,000 (Direct Cost: ¥4,200,000)
Fiscal Year 2002: ¥1,400,000 (Direct Cost: ¥1,400,000)
Fiscal Year 2001: ¥2,800,000 (Direct Cost: ¥2,800,000)
|
|---|
| Keywords | line drawing / depth perception / model / entropy / simulation |
|---|
| Research Abstract |
Line drawings are more often used in the documents that illustrate the functions and the shapes of products or buildings rather than the photo-realistic images created by computer graphics. For example the figures in the manual of a personal computer are mostly line drawings. It is not the reason of it that line drawings are more economical of production costs than the photo-realistic images, but line drawings are more suitable to convey the shapes of 3-D objects than the photo-realistic images. It is important to understand how to interpret 2-D line drawings as 3-D objects for communication about the shapes of 3-D objects in man-machine interface.
First, we proposed a principle for 3-D perception from single line drawings motivated by T. Marill's principle (1992) to minimize the description length. The proposed principle is to minimize the entropy of angle distribution (MEAD) between line segments in a 3-D wire frame, whose 2-D projection coincides with the given line drawings.
Secondly, we implemented the principle MEAD using a genetic algorithm as a simulation program. The results of simulation experiments show that the proposed principle of MEAD provides 3-D wire frames similar to the 3-D perception from the given 2-D line drawings.
Finally, we proposed a model of 3-D perception for simple closed curves included in line drawings as ridgelines of 3-D objects. The 3-D perception of closed ridgelines can be roughly divided into two appearances. We defined two evaluation functions based on the two appearances, and implemented them as a simulation program. The results of simulation experiments show that the model provides 3-D curves similar to the 3-D perceived ones from the given 2-D ridge lines.
|
