Abstract: In this paper, an improved method based on digital change detection of multi-temporal high resolution RS images is proposed to extract the building damage caused by catastrophic earthquakes, which aims to improve the accuracy of classification and to reduce the impact of the deformation of corresponding image point caused by low registration precision, different photographic mode, and the differences of surface elevation and height of buildings by searching the maximum image correlation coefficient of a point on post-event image from a certain spatial domain around the registration point on the pre-event image instead of the conventional correlation coefficient exactly on the the registration point. The improved method is applied to detect the changes of buildings in a test area of the city of Dujiangyan after the 2008 Wenchuan earthquake by using pre-earthquake QuickBird remote sensing data acquired on July 22, 2005 and post-earthquake aerial images acquired on May 18, 2008. The results of the experiments are compared with the results using traditional method and indicate that the improved method is more precise and stable to extract the seismic damage information than traditional method.

Key words: Correlation coefficient, Change detection, Seismic damage extraction of buildings, Regional searching optimization method