Airborne laser scanning: clustering in raw data
Marco Roggero · 2001
We implemented a strategy for terrain, vegetation and building detection, based on laser range data only. The result was obtained by working on raw data, so we were able to take advantage of the full resolution potential of laser scanning. The detection of objects was performed in two stages: first, elevated objects and ground are separated, and then the objects are classified as vegetation or buildings. Work is still in progress, about the extraction and classification of entities. A comparative analysis of the first pulse, the last pulse and intensity data can improve the result of clustering. Results obtained in different environments with one-meter grid laser data are shown; we have tested the algorithm on city areas, countryside, river bed, landslides, mountains and wooded terrain. 1