Automatic building edge extraction from lidar data based on images segmentation

LIDAR (Light Detection And Ranging) has become a reliable technique for the data collection of the earth surface. Building extraction based on LIDAR data is one of the important applications. The image edge is one important feature of a building, which can reflect the most valuable information of the image. And it can roughly describe the characteristics of the building. So building edge extraction from LIDAR data is of great significance. This paper presents an automatic extraction method of building edges from LIDAR data, using image segmentation technology. In the first stage, the LIDAR point cloud data are converted into DSM images. In the second stage, a threshold selection method based on the transition region determination is used to segment the DSM images. Finally, the building edges are extracted on the segmented image. The results obtained on the data sets with different ground point density, gathered over the town of Xuzhou (China) with TopScan airborne laser scanning system, are shown to illustrate the effectiveness of the proposed approach.