Determining Building Footprints

A new approach for determining building boundaries through automatic processing of light detection and ranging (LIDAR) data is presented. In general, LIDAR data is processed and interpolated into grayscale images with pixel values corresponding to height measurements. In a recent work, a progressive morphological filter was used to separate ground from non-ground measurements. Further separation of non-ground measurements into building and non-building measurements was performed by growing regions with similar characteristics. The initially identified building outlines were then refined using a polygonal fitting algorithm to obtain ground plan building representations known as building footprints. In this paper, the building outline refinement process is improved through a clean-up and sorting procedure, in order to provide more accurate building footprints.