3D city models: Mapping approach using LIDAR technology

Abstract

We propose a map building system for generation of 3D city models using a vehicle equipped with LIDAR (Light Detection And Ranging) technology, which is capable of obtaining more than one million points per second. This approach is based in two methods: plane segmentation and local map construction. The First method introduces a fast method to extract the ground plane. The rest of the points are used to extract planes using a general plane extraction. For the general plane extraction process, is used a method that choose a random point and its neighbor in order to compute a plane; this plane is evaluated with respect to the inliers number, and the best plane is the one containing the biggest number of points. Finally, using the planes, we construct an environment map.