1996 Fiscal Year Final Research Report Summary
Development of Analysis Processing System Using Orbital Satellite Image Data
| Project/Area Number |
07558271
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 試験 |
|---|
| Research Field |
情報システム学(含情報図書館学)
|
|---|
| Research Institution | Ibaraki University |
|---|
Principal Investigator |
HOSHI Takashi Faculty of Engineering, Ibaraki University, Professor, 工学部, 教授 (80026129)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TONOOKA Hideyuki Faculty of Engineering, Ibaraki University, Assistant, 工学部, 助手 (80261741)
TORII Kiyoshi Faculty of Agriculture, Kyoto University, Associate Professor, 農学部, 助教授 (40026563)
SATO Kazuhiro Faculty of Agriculture, Ryuku University, Associate Professor, 農学部, 助教授 (50045126)
|
|---|
| Project Period (FY) |
1995 – 1996
|
| Keywords | Remote Sensing / Image Processing System / Orbital Satellite Image / Geometric Transformation / K-means / Fuzzy-Neural Network / Verification of Training Area Data |
|---|
| Research Abstract |
The SAIAS system that processes image data extracted from passive sensors onboard in orbital satellites is improved and renovated. The contents of this innovated IBASYS system is described below :
(1) The original FORTRAN programs are translated to C++, making possible to run on workstations and personal computers.
(2) To verify the data conditions necessary for analysis preparation, the color pattern is expanded from 7 to 18 bands.
(3) To improve the processing efficiency of geometric transformation of an image, the man-machine processing of error analysis and the image transformation processing was hierarchically added.
(4) The K-means method is added in the unsupervised classifications, developing an algorithm using a raster classification processing.
(5) A general notion algorithm to verify the application of a selected training data on an image data is added.
(6) As an expansion to the classification methods, a theoretical investigation is made on FNN classification, a combination of fuzzy theory and neural network theory, and is made possible to classify an orbital satellite image data.
|
Research Products
(12 results)
