| Project/Area Number |
61850064
|
|---|
| Research Category |
Grant-in-Aid for Developmental Scientific Research
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
計算機工学
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
TSUJI Saburo Professor, Faculty of Engineering Scienece, oSaka University, 基礎工学部, 教授 (60029527)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
UCHIYAMA Takashi Project Leader, Control Engineering Laboratory, Fujitsu Laboratories, 制御研究所, 室長
ASADA Minoru Assistant Professor, Faculty of Engineering, Osaka University, 工学部, 講師 (60151031)
ABE Norihiro ibid Associatr Professor, Faculty of Engineering Scince, Osaka University ibid, 基礎工学部, 助教授 (00029571)
YACHIDA Masahiko Associate professor, fuculty of enginnering Scince, Osaka University, 基礎工学部, 助教授 (20029531)
|
|---|
| Project Period (FY) |
1986 – 1987
|
| Project Status |
Completed (Fiscal Year 1987)
|
| Budget Amount *help |
¥15,200,000 (Direct Cost: ¥15,200,000)
Fiscal Year 1987: ¥7,700,000 (Direct Cost: ¥7,700,000)
Fiscal Year 1986: ¥7,500,000 (Direct Cost: ¥7,500,000)
|
|---|
| Keywords | Stereo / time varyin imagery / image processor / three-fimensional / 世界モデル / 画像プロセッサ |
|---|
| Research Abstract |
The objective of the research is to establish system design ot real time 3-D wivion which acquires 3-D information from multiple images. A system simulating the 3-D vision was built by connercting a high-speed image processor whrh SUM3 workstations, in order to test performatnces of vatious algorithms. We have developed the following mwthods:
(1) a high speed segment-based stereo which finds correspondence between segment reliably and efficiently, (2) a trinoculat stereo which discatds false correspoundece betweeen sdge points using geometriacal constraints, (3)a stereo forma mobile robsts which uses constraints of indoor scenes for efficient establishing of correpsondence,
(4) a real time method for acquiringd 3-D shapge by projecting a stripe pattern,
(5) an active stereo method which finds correspondence between regions in stereo images of a scene projected a tandom dot partern, (6) a high speed object recognition system in which many processors vote onto multi-dimensional accumulator using a genralized Housh method,
(7) a system which maps image data acquired at different locxations onto an objecxt-centrred coordinate-system and builds a worls model.pErformance and featrures of the above-mentioned system are studed and the design concept of the 3-D vision system is established.
|
