Reconstruction of building ground plans from laser scanner data. In: AGILE08

Abstract. Laser scanner data are a powerful data source for acquisition and update of high resolution topographic information. Current airborne laser scanning systems are able to acquire densities of up to 40 points per square meter. With this information topographic objects like buildings, trees, electricity lines, and the relief can be determined. The main problem is the classification and identification of these objects in the point cloud and their precise reconstruction. In this paper we tackle the problem of delineating building ground plans from laser scanner point clouds. Starting point is a pre-processing step that distinguishes terrain and off-terrain points. Then, all connected off-terrain points are aggregated in a region growing process. These regions represent potential buildings. For them, in a third step, the outline is simplified. This is a generalization task, which has to take the characteristics of buildings into account to produce a meaningful 2D building shape. In the paper, these steps are described in detail. The focus lies on the different possibilities to generalize the building ground plans. We describe options for this generalization and especially focus of an approach using RANSAC followed by least squares adjustment. INTRODUCTION AND STATE OF THE ART Airborne laser scanner data describe the top surface of objects in terms of a point cloud in grea