會議論文
| 學年 | 100 |
|---|---|
| 學期 | 1 |
| 發表日期 | 2011-12-05 |
| 作品名稱 | Flood Identification from Satellite Using Neural Networks |
| 作品名稱（其他語言） | |
| 著者 | Chang L. C.; Kao I. F.; Shih K. K. |
| 作品所屬單位 | |
| 出版者 | |
| 會議名稱 | 2011 AGU Fall Meeting |
| 會議地點 | San Francisco, USA |
| 摘要 | Typhoons and storms hit Taiwan several times every year and they cause serious flood disasters. Because the rivers are short and steep, and their flows are relatively fast with floods lasting only few hours and usually less than one day. Flood identification can provide the flood disaster and extent information to disaster assistance and recovery centers. Due to the factors of the weather, it is not suitable for aircraft or traditional multispectral satellite; hence, the most appropriate way for investigating flooding extent is to use Synthetic Aperture Radar (SAR) satellite. In this study, back-propagation neural network (BPNN) model and multivariate linear regression (MLR) model are built to identify the flooding extent from SAR satellite images. The input variables of the BPNN model are Radar Cross Section (RCS) value and mean of the pixel, standard deviation, minimum and maximum of RCS values among its adjacent 3×3 pixels. The MLR model uses two images of the non-flooding and flooding periods, and The inputs are the difference between the RCS values of two images and the variances among its adjacent 3×3 pixels. The results show that the BPNN model can perform much better than the MLR model. The correct percentages are more than 80% and 73% in training and testing data, respectively. Many misidentified areas are very fragmented and unrelated. In order to reinforce the correct percentage, morphological image analysis is used to modify the outputs of these identification models. Through morphological operations, most of the small, fragmented and misidentified areas can be correctly assigned to flooding or non-flooding areas. The final results show that the flood identification of satellite images has been improved a lot and the correct percentages increases up to more than 90%. |
| 關鍵字 | |
| 語言 | en_US |
| 收錄於 | |
| 會議性質 | 國際 |
| 校內研討會地點 | 無 |
| 研討會時間 | 20111205~20111209 |
| 通訊作者 | Li-Chiu Chang |
| 國別 | USA |
| 公開徵稿 | |
| 出版型式 | |
| 出處 | 2011 AGU Fall Meeting |
| 相關連結 |
機構典藏連結 ( http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/106699 ) |