Damage Detection Based on Object-based Segmentation and Classification from High-resolution Satellite Images for the 2003 Boumerdes, Algeria Earthquake

A strong earthquake of magnitude 6.8 struck the Mediterranean coast of Algeria on 21 May 2003 and the city of Zemmouri in Boumerdes province was most heavily damaged. QuickBird satellite observed the Zemmouri area on 23 May 2003. By image sharpening, buildings, cars and even debris can clearly be identified in a natural colour image. Preliminarily, the present authors performed visual damage inspection comparing the post-event image with an image acquired before the earthquake. As a result, totally collapsed buildings, partially collapsed buildings, and buildings surrounded by debris were visually identified. Additionally, debris surrounding damaged buildings was also extracted. Although these observations indicate that high-resolution satellite images would be able to provide quite useful information to emergency management after natural disasters, it can also be said that the visual damage interpretation is time-consuming. For practical purposes, it must be necessary to complete damage detection as quick as possible after the occurrence of disasters in order to make use of the detection result in emergency management. Hence, an automated damage detection method, in which debris is identified, is required to be developed. In this study object-based image segmentation and classification technique as well as pixel-based technique have been applied. This technique would make it possible to consider not only the spectral characteristics of objects but also the spatial relationship between objects that consist of homogenous pixels. For the purpose of investigating their effectiveness on identifying debris, the accuracy of the detection result has been assessed and compared with that of the pixel-based damage detection result.