Building Detection and Description from a Single Intensity Image

We describe a method to detect buildings and construct 3-D shape descriptions of the buildings from a monocular aerial image of general viewpoint. A grouping process is used to generate 2-D roof hypotheses from fragmented linear features extracted from the input image. Good hypotheses are selected based on 2-D evidence and some local 3-D evidence. These are then verified by searching for 3-D evidence consisting of shadows cast by the roof and walls associated with it. This process also helps construct 3-D models of the verified buildings. Overlap and containment relations between 3-D structures are analyzed to resolve conflicts. The method also allows for integration of results obtained from multiple images. This system has been tested on a large number of real examples with good result, some of which, and their evaluation, are included in the paper.