Abstract
In order to develop an automatic damage detection technique, the characteristics of texture in high-resolution satellite images observed before and after the 2010 Haiti earthquake are examined. From the texture analysis based on gray-level co-occurrence matrix, the dissimilarity of the images is identified as better classifier than other texture indices in detecting collapsed buildings. By using the dissimilarity calculated from the pre- and post-event images, damage detection is performed to identify the distribution of the collapsed buildings and the accuracy of the detection is assessed by the ROC analysis. The result shows that about 70% of the collapsed buildings are correctly detected by the proposed method.
