A Determination of the Earthquake Disaster Area by Object-based Analysis Using a Single Satellite Image

Abstract

When a large-scale natural disaster occurs, it is important to grasp an overall picture of the damage that the area has suffered as soon as possible. In addition, maps and images that show the status and location of the damage in the area are necessary to support the efficiency of emergency relief services such as firefighting, volunteers and other groups. However; the wider an affected area, the more time it will take to confirm the damage, if such efforts and processes rely on human power. This challenge should be addressed technologically by developing a method capable of analyzing an affected area within a short period of time so as to provide useful information for emergency relief and rescue operations. The final goal of this study is to provide data for supporting emergency relief efforts in a disaster affected area by locating damaged buildings shortly after the disaster. In this study, the importance of time in emergency situations is prioritized by designing a method that only uses a single satellite image of an affected area, eliminating the use of complex algorithms and auxiliary data. The uniqueness of our method lies in the application of object-based region segmentation to images and the use of features of objects obtained from texture, hierarchical and other information in order to extract damaged buildings. Out of 26 features resulting from the analysis of objects, we found one feature and three combinations of two different features that are effective in extracting damaged buildings, such as Rectangular fit, Homogeneity, Number of sub-objects/Area, and Length of longest of edge/Area.