| With the deeply application of 3D GIS and virtual realism, conventional measuring methods such as engineering survey and photogrammetry were not seasoned with some fields because of its'scarcity in work efficiency and data precision. Airborne LIDAR (Light Detect And Ranging) is a new measuring tool which is being widely used in many different kinds of applications. It explores objects and measures distance by the way of laser scanning and can produce urban digital surface model (DSM) in a fast rapid. A new way for urban building extraction was occurred when the technology was succeeded.In this paper, we investigate the urban building extraction from airborne LIDAR points cloud data. First, the massive discrete LIDAR points cloud data is regular grid re-sampled and translated into DSM, then the DSM is gray quantized to a DSM depth-image according to its elevation. Secondly we preprocess the DSM depth-image using image preprocessing technique. Using histogram threshold segmentation of DSM depth-image ground area is wiped off. The small area objects are removed by tab area measurement method and the interference information is filtered, the independence building area is gained. The last step is extraction of the urban building. Using Canny edge detector to do with the building edges. After then apply improved Hough transformation to the edges. The improved Hough transformation takes full use of data of Hough parameter space, and get the coordinates of two parallel line segments'start and end point in the rectangle building. After based on rectangle building edges'vertical and parallel conditions, we calculate buildings'four corner points coordinates, and the normalization of rectangle building edges can be completed. And then the extraction of the rectangle building in DSM depth-image is accomplished.In a word, this paper based on the airborne LIDAR points cloud data achieves the extraction of urban building by using digital image processing technique without any assistance data. |
PDF Full Text Request