A semi-automated approach for extracting buildings from QuickBird imagery applied to informal settlement mapping

Recent advances in sensor technology have promoted the mapping communities to investigate the potential and information contents of recent very high-resolution satellite images. In this paper, we report our new semi-automatic building extraction approach and our first results of mapping informal settlement areas obtained using QuickBird high-resolution images. We implemented our mapping approach using snakes and a radial casting algorithm, and assessed the results both qualitatively and quantitatively and compared them with ground truth data from a similar area. Finally, we summarized the potential and limitations of the second-generation commercial high-resolution satellite images to extract buildings using existing software.