2001 Fiscal Year Final Research Report Summary
Development of multi-eyes camera by using smart image sensors and its application to high speed image processing system
| Project/Area Number |
12838010
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
複合化集積システム
|
|---|
| Research Institution | Tokyo University of Science |
|---|
Principal Investigator |
HAMAMOTO Takayuki Tokyo University of Science, Dept. of Electrical Eng., Lecturer, 工学部, 講師 (10297624)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
AIZAWA Kiyoharu University of Tokyo, Dept. of Frontier Info., Professor, 新領域創成科学研究科, 教授 (20192453)
|
|---|
| Project Period (FY) |
2000 – 2001
|
| Keywords | Computational sensor / Smart image sensor / Vision chip / Image processing system / Real time processing / FPGA |
|---|
| Research Abstract |
In this research, we have investigated a new computational image sensor which has image processing circuits on a same chip and its application to high speed image processing system.
1 We have investigated the algorithm for on-chip or off-chip image processing. The sensor detects the difference between two frames and outputs the position information of the detected pixels.
2 We have designed a prototype LSI of CMOS image sensor which has 128x128 pixels. It can operate at 5000 frame / second. The sensor was designed by column parallel architecture and the image processing can be done by the analog processing circuits. We have fabricated the LSI chip by the fabrication service CMP.
3 We have fabricated the binocular stereo camera system by using the smart sensor chip.
4 We have designed the high speed image processing system by using the camera system and FPGA chip which can process the pixel value and position information of the moving pixels in real time. It can operate at more than 1200 frame / second.
5 We have verified that the several moving objects can be tracked by using the processing system in parallel. Furthermore, the system can estimate the depth between the objects and the camera in real time. We can get 15cm accuracy of depth estimation at high frame rate.
|
