Simple building reconstruction from LIDAR point cloud

This paper presents a method for simple regular building reconstruction from LIDAR point cloud. At First, TIN model of the extracted building point from discrete LIDAR point cloud is built. Those points in triangle facet which having the similar normal vector value are clustered into the same plane point set. Then, every plane point set is fitted into plane using the least square method which boundary is decided by convex hull theory. The boundary of building top plane can be optimized based on regular building principle. From the optimized building top boundary and the height of ground, we can calculate the 3D coordinate of every corner point on building. The building model can be constructed based on 3D coordinates of these corner points. The experimental result shows that this method is able to construct effectively simple regular building.