Extraction of dense urban area from high resolutions satellite images

In recent years, we have come to be able to use very high resolution images from satellites. The images of urban areas are useful for prediction of damages by disaster or for maintenance of map data. For disaster prevention, especially in dense urban area, we need to accurately know the density of buildings. In this paper, we propose a method which extracts dense urban areas automatically from high resolution satellite images. In this research, we use high resolution QuickBird images for experiments. The method first extracts candidate region of dense urban area from a panchromatic image and extracts vegetation areas and water areas from a multi-spectral image. Then the method finally extracts dense urban areas by combining the information. We verified that the method can extract the dense urban areas appropriately.