Preliminary investigations into feature extraction from LIDAR elevation and intensity data have produced 3-D building footprints, along with other 3-D data layers of trees and forest polygons, powerline strikes, and bare-earth LIDAR DEM surfaces. The feature extraction evaluation has focused on the Rapid Terrain Visualization (RTV) LIDAR Toolkit, an ArcView-based extension. LIDAR data from several vendors were tested using the toolkit, with good building-extraction results from suburban areas, but inadequate extraction in high-density urban core areas containing oversize or odd-shaped buildings. Successfully extracted buildings and bare-earth DEM data produced by this toolkit have been fed into the CATS (Consequences Assessment Tool Set) software as part of the high-resolution data required for Homeland Security threat assessment.

Cooperative agreements both within academia and across federal agencies are also being pursued to provide the USGS with unlimited access to other feature extraction methods and software packages. Emphasis is currently focused on the extraction of 2-D and 3-D building footprints in support of the National Map and Homeland Security efforts, but extraction of other important data layers such as bare-earth elevation models, roads, trees, and hydrography is under investigation as well. 

The image below shows a screen capture of the Arcview extension, RTVToolkit, showing building outlines, in red, extracted from the color-coded LIDAR data:

The next image below shows a screen capture of how LIDAR data can be used in modeling programs. The LIDAR data is integrated into the CATS to simulate the airflow around buildings when Sarin gas is released into an urban area.

Project Lead: John Kosovich