2000 Fiscal Year Final Research Report Summary
Development of hierarchical software for the processing of remotely sensed multi-dimensional images to be used for the evaluation of global environment
| Project/Area Number |
10555135
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B).
|
| Allocation Type | Single-year Grants |
|---|
| Section | 展開研究 |
|---|
| Research Field |
Measurement engineering
|
|---|
| Research Institution | Teikyo Heisei University (2000)
The University of Tokyo (1998-1999) |
|---|
Principal Investigator |
SADAO Fujimura Teikyo Heisei University, Faculty of Informatics, Professor, 情報学部, 教授 (30010961)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HANAIZUMI Hiroshi Hosei University, Faculty of Engineering, Professor, 工学部, 教授 (60143385)
KUSAKA Takashi Kanazawa Institute of Technology, Faculty of Engineering, Professor, 工学部, 教授 (20064454)
KIYASU Senya Teikyo Heisei University, Faculty of Informatics, Lecturer, 情報学部, 講師 (20234388)
IWATA Yoshitaka Fujitsu FIP Corporation, Research Engineer, 環境システム事業推進部, 研究員
MIZOTE Muneaki Teikyo Heisei University, Fauclty of Informatics, Professor, 情報学部, 教授 (70009645)
|
|---|
| Project Period (FY) |
1998 – 2000
|
| Keywords | remote sensing / multispectral image / hierarchical processing / global environment |
|---|
| Research Abstract |
The purpose of this research was twofold. The first one is to develop new algorithms for hierarchical processing of remotely sensed multispectral data. The other is to make a software package including the hierarchical processing algorithms for the images above..
In the first one, we aimed at extending the functions of the algorithms we had already developed, and also developing new algorithms. We refined algorithms for automatic registration to make highly efficient one. The algorithm is to register multispectral images by dividing images into small triangles and by transforming then by a linear transform to match the triangles with each other. We developed also new classification algorithms. In one of them we devised a method to estimate a priori probability for each category using spatially hierarchical structure of the image to achieve high accuracy. We also succeeded high accuracy and high efficiency at the same time by developing purpose-oriented feature extraction algorithm.
In the realization of software package, we discussed to finalize the system structure of the package, unifying the specification of input and output image data format, the interface among the algorithms and user interface. We included several algorithms into the package which are "automatic registration algorithm, ""un-supervised classification (clustering) algorithm, " and "binary decision tree classification algorithm."
|
