An Improved Building Detection Technique for Complex Scenes

The success of automatic building detection techniques lies in the effective separation of buildings from trees. This paper presents an improved automatic building detection technique that achieves more effective separation of buildings from trees. Firstly, it uses cues such as height to remove objects of low height such as bushes, and width to exclude trees with small horizontal coverage. The height threshold is also used to generate a ground mask where buildings are found to be more separable than in a so-called normalized DSM (digital surface model). Secondly, image entropy and colour information are jointly applied to remove easily distinguishable trees. Finally, an innovative rule-based procedure is employed using the edge orientation histogram from the imagery to eliminate false positive candidates. While tested on a number of scenes from four different test areas, the improved algorithm performed well even in complex scenes which are hilly and densely vegetated.