Share Email Print

Proceedings Paper

The method of earthquake landslide information extraction with high-resolution remote sensing
Author(s): Jian Wu; Peng Chen; Yaolin Liu; Jing Wang
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

As a kind of secondary geological disaster caused by strong earthquake, the earthquake-induced landslide has drawn much attention in the world due to the severe hazard. The high-resolution remote sensing, as a new technology for investigation and monitoring, has been widely applied in landslide susceptibility and hazard mapping. The Ms 8.0 Wenchuan earthquake, occurred on 12 May 2008, caused many buildings collapse and half million people be injured. Meanwhile, damage caused by earthquake-induced landslides, collapse and debris flow became the major part of total losses. By analyzing the property of the Zipingpu landslide occurred in the Wenchuan earthquake, the present study advanced a quick-and-effective way for landslide extraction based on NDVI and slope information, and the results were validated with pixel-oriented and object-oriented methods. The main advantage of the idea lies in the fact that it doesn’t need much professional knowledge and data such as crustal movement, geological structure, fractured zone, etc. and the researchers can provide the landslide monitoring information for earthquake relief as soon as possible. In pixel-oriented way, the NDVI-differential image as well as slope image was analyzed and segmented to extract the landslide information. When it comes to object-oriented method, the multi-scale segmentation algorithm was applied in order to build up three-layer hierarchy. The spectral, textural, shape, location and contextual information of individual object classes, and GLCM (Grey Level Concurrence Matrix homogeneity, shape index etc. were extracted and used to establish the fuzzy decision rule system of each layer for earthquake landslide extraction. Comparison of the results generated from the two methods, showed that the object-oriented method could successfully avoid the phenomenon of NDVI-differential bright noise caused by the spectral diversity of high-resolution remote sensing data and achieved better result with an overall accuracy of 92.16%, while of the pixel-oriented one could only get 71.32%. As the high-resolution remote sensing has been widely utilized in many fields, the object-oriented image analytical technique will have an extensive application.

Paper Details

Date Published: 14 May 2014
PDF: 6 pages
Proc. SPIE 9158, Remote Sensing of the Environment: 18th National Symposium on Remote Sensing of China, 915805 (14 May 2014); doi: 10.1117/12.2063691
Show Author Affiliations
Jian Wu, Third Institute of Oceanography, State Oceanic Administration (China)
Peng Chen, Third Institute of Oceanography, State Oceanic Administration (China)
Yaolin Liu, Wuhan Univ. (China)
Jing Wang, Key Lab. of Land Use, Ministry of Land and Resources (China)

Published in SPIE Proceedings Vol. 9158:
Remote Sensing of the Environment: 18th National Symposium on Remote Sensing of China
Qingxi Tong; Jie Shan; Boqin Zhu, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?