1990 Fiscal Year Final Research Report Summary
Feature extraction of eye and lip movements from video image of the face
| Project/Area Number |
63460129
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
電子通信系統工学
|
|---|
| Research Institution | Kanazawa Institute of Technology |
|---|
Principal Investigator |
KAKITA Yuki Kanazawa Institute of Technology, Faculty of Engineering, Professor, 工学部, 教授 (00098823)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HONDA Kiyoshi Kanazawa Institute of Technology, Faculty of Engineering, Associate Professor, 工学部, 助教授 (10114638)
KAWARA Tetsuo Kanazawa Institute of Technology, Faculty of Engineering, Professor, 工学部, 教授 (40112776)
SHIMADA Youichi Kanazawa Institute of Technology, Faculty of Engineering, Professor, 工学部, 教授 (50113155)
|
|---|
| Project Period (FY) |
1988 – 1990
|
| Keywords | Communication / Face posture / Eye movement / Lip movement / Image Processing / Speech / Language / Human Interface |
|---|
| Research Abstract |
A. System development
A group of computers and peripheral devices related to the image, sound and bio-electric signals, were facilitated to our research group by the aid from Monbusho for special research facility in 1987 (Showa 62). We developed the additional system for this particular research project. Newly purchased facilities by the aid of this project were a workstation for image processing, image capture board, and RAMs. Completed systems were good for our research purpose.
B. Preliminary Results
Pilot experiments were done for the extraction of lip movements and eye movements.
(1) eye movements
Experiments of detecting the eye pointing object were conducted. The image system estimate the direction of eye sight by the image of iris of one eye when looking at the 5x5 objects configured in matrix. Under the head still condition, good correlation was obtained between the position of object and the image parameters of iris image. For head free condition, a preliminary method, by applying a special coordinate conversion, was examined. It was also promising.
(2) lip movements
Using 3 parameters, vertical opening of lips and upper-lower incisors, and horizontal opening, differentiation of 11 consonants in the context of /aCa/. Almost all the consonants, except /k,g/ and /r/, were differentiated successfully. For the error of recognition was suspected to be caused by the low frame rate. A new image interpolation method were developed. The method is promising because it can enhance the time resolution substantially.
|
