2023 Fiscal Year Final Research Report
Automatic extraction of landslide areas by machine learning of satellite data using deep learning
| Project/Area Number |
20K05054
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 25030:Disaster prevention engineering-related
|
|---|
| Research Institution | Hiroshima Institute of Technology |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2020-04-01 – 2024-03-31
|
| Keywords | 合成開口レーダ / 畳み込みニューラルネットワーク / 土砂崩壊 |
|---|
| Outline of Final Research Achievements |
In order to understand the damage situation at an early stage, landslide areas were extracted using a full-layer convolutional neural network based on synthetic aperture radar (SAR) data onboard a satellite. We also extracted landslide areas using optical sensor data and compared the extraction accuracy. Using test data generated from X-band and L-band SAR data pre- and post-event The 2018 Hokkaido Eastern Iburi Earthquake, the optimal values of parameters for landslide area extraction were evaluated using F-values, and their effectiveness was demonstrated.
|
| Free Research Field |
リモートセンシング
|
| Academic Significance and Societal Importance of the Research Achievements |
広域災害等において被災状況を早期把握するためには全天候型センサであるSARの活用が有効であるが、幾何学的な歪などにより目視による土砂崩壊地の抽出は容易ではない。本研究では画像認識で成果を上げている深層学習を用いて災害前後のXバンドおよびLバンドSARデータから土砂崩壊地抽出を行い、検証用データとの比較から土砂崩壊地抽出の精度を示した。これにより災害発生時におけるSARデータの利用可能性を示すことができた。
|
