posted: April 23, 2013 | author: Woolpert Labs
Provided by guest blogger, Daniel Ngoroi, Woolpert subject matter expert, Dayton
The main challenge posed by the ubiquity of LiDAR data is the construction of realistic polygonal models from the unstructured and random point cloud. There are many methods, tools and techniques for constructing surfaces from 3D point clouds, which are often bundled along with other functions in large software packages.
Woolpert has long sought a targeted solution to the problem of LiDAR exploitation, so we devised and developed a framework, Automated Building Extraction (ABE), that uses open-source data toolkits and libraries to detect objects in LiDAR data, extract them and construct realistic surfaces representing solids within the point cloud. The main goal of this approach was to create a framework that extracts accurate polygons for use as a mapping product in GIS from LiDAR data of varying densities.
Toolkits such as the Point Cloud Library, GDAL, PDAL and libLAS were incorporated in the framework to overcome the challenges of unstructured point clouds and extract planes and edges of physical structures. Open-source tools are highly customizable and enable the detailed investigation of point cloud characteristics and manipulation of modeling parameters for the production of robust solutions.
This framework has been successfully used on multiple projects to extract building footprint polygons and derive information that could be used in other mapping applications.