Complex building roof detection and strict description from LiDAR data and orthorectified aerial imagery

One common and important task in urban modeling is the extraction of 3D building geometry. Detecting and describing rooftops is a key step. We propose a building roof detection and description method using airborne LiDAR data and orthorectified aerial imagery. Our approach makes no presumption of the shape that a building rooftop should exhibit. We use LiDAR to separate the building roof and its other significant features from other objects in the scene and estimate the strict outlines of these features. Imagery is processed using a robust segmentation and edge detection method to generate an edge map of the roof and then used to help refine the outlines obtained from LiDAR data. We will illustrate our approach by extracting several roof boundaries in the downtown region of Rochester, NY that exhibit simple and complex outlines.