| Project/Area Number |
07558034
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 試験 |
|---|
| Research Field |
計算機科学
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
SAKAUCHI Masao Institution of Industrial Science The University of Tokyo Prof., 生産技術研究所, 教授 (30107370)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YAGINUMA Yoshitomo Institution of Industrial Science The University of Tokyo Research Associate, 生産技術研究所, 助手 (10251464)
|
|---|
| Project Period (FY) |
1995 – 1996
|
| Project Status |
Completed (Fiscal Year 1996)
|
| Budget Amount *help |
¥10,900,000 (Direct Cost: ¥10,900,000)
Fiscal Year 1996: ¥4,000,000 (Direct Cost: ¥4,000,000)
Fiscal Year 1995: ¥6,900,000 (Direct Cost: ¥6,900,000)
|
|---|
| Keywords | multimedia / database / hypermedia / digital broadcasting / video recognition / リアルタイムリンク / マルチメディアプラットホーム / 放送映像自動認識 / 映像シーン記述言語 |
|---|
| Research Abstract |
The objective of this research project is to develop a new type of hypermedia system with capability of advanced realtime linking to broadcasted video data.
Major results of the project are as follows.
(1) The Live Hypermedia System, which can acquire a desired set of video scenes from live television broadcasting programs using automated video scene recognition techniq, ues has been originally developed. In the system, information nodes and links are dynamically determined corresponding to the structure of the recognition model. The recognition models itself are expressed by the flexible compounds of state-transition rules and scene feature descriptions.
This Live Hyper-media System is implemented on our original platiorm GOLS.In order to demonstrate the effectiveness, experimental evaluations of the Recall rate and Precision rate (video scence pick-up performance) for TV broadcasting programs have been done successfully.
(2) To realize more flexible and user-oriented access and utilization of video data, more advanced methods of making high data models have been developed. Basic idea is to realize logical synchronization between related video, sound and document data.Higher-level recognition, retrieval and date authorization are successfully developed based on this idea. Experimented results reveal the effectiveness of there methods.
|
