Automated extraction of building geometric features from raw LiDAR data

In recent years, with new services expected, such as navigation systems, location based services, and augmented reality, the need for automatically, efficient 3D building reconstruction systems becomes more urgent than ever. As a new spatial information technology, airborne LiDAR is widely used for the acquisition of 3D objects on the earth. The automatic reconstruction of 3D buildings from airborne LiDAR data has been a topic of research for decades. However the lack of 3D building reconstruction methods is still the bottleneck for the further development of airborne LiDAR. Since the 3D geometry feature is extremely important for many applications such as urban planning, car navigation, or environment monitoring etc., this paper introduces an automated method for implementing building geometric features extraction from raw LiDAR data in details. The outlines of 3D buildings are generated by using discrete curvature analysis according to the building geometry features. In the experiment, 3D building models are automatically reconstructed by using a model-driven approach according to their geometry features derived from raw LiDAR data.