Earthquake building damage detection with object-oriented change detection

Remote sensing image change detection is an effective method to estimate earthquake damage. With the improvement of spaceborne image resolution, single pixel is gradually replaced by object as the basic unit for change detection. The principle of object-oriented change detection method is described in this paper. And then data processing flow is stated. Taking high-resolution remote sensing images pre- and post- the Yushu earthquake as an example, the accuracy of this method to extract building damage is shown. This method is compared with direct change detection method based on pixel, and based on principal component analysis. The results show object-oriented change detection method has the highest accuracy in building damage estimation, which is 83%, while the accuracy of the latter two methods are 72% and 61%.