One of the more compelling new technologies at the SPAR 2009 conference is software from ClearEdge3D called EdgeWise that automatically extracts planar surfaces from point clouds to create a 3D CAD model. The software harnesses computer vision algorithms to pull coplanar points from point cloud clusters to create surfaces that can then be exported in DXF format to CAD software such as AutoCAD, MicroStation and SketchUp. The resulting model needs some clean up, but clearly solves the time-consuming and mostly manual task of creating building models from terrestrial LIDAR data, while preserving and even enhancing the accuracy of the LIDAR data.
Kevin Williams, CEO of ClearEdge3D described the process to me as an ingestion of point clouds taken from different angles to triangulate to create polygons. Rather than creating a large LIDAR mesh with a large number of polygons that can be overwhelming given that the files are many times larger than an original scan, the EdgeWise polygon extraction pulls just the surface polygons resulting in a small file size. ClearEdge3D produces a precise model that enhances accuracy given the multiple perspectives.
The software even has the capability to automatically classify different surface regions that can be color coded in order to ease the final model creation process or for inspection purposes. The resulting model can also include an optimized ground surface model, making it an excellent fit for GIS applications.
This exciting new technology tackles one of the primary pain points of 3D model creation from LIDAR points, and has some compelling prospects to help bridge the CAD/GIS and BIM modeling worlds.