Study On The Building Extraction From Airborn LIDAR Points Cloud Data

In the end of last century, airborn LIDAR technology has made a breakthrough inreal-time collection of three-dimensional geospatial information. It provides a new technicalmethod for the acquisition of high spatial-temporal resolution geo-spatial information.Because of acquiring three-dimensional coordinates automatically and accurately, andprotecting from the weather, this technology makes a transition from traditional single dataacquisition manually into a continuous data acquisition automatically, which greatly liberatesthe human productive forces. As the result, the automation and intellectualization of datapost-processing become a significant problem to solve.This paper is based on airborn LIDAR point cloud data and used some arithmetic to fillthe data hole, then gives regular grid re-sampling to the irregular discrete LIDAR point clouddata,which result in digital surface model (DSM).Then,DSM depth image are generated byquantify the gray scale in accordance with elevation. After that, using histogram thresholdsegmentation and small area removing method to preprocess the DSM depth image, theindependent building area are gained. Finally, using the rough edge detection of buildingCanny edge detection operator, line group to obtain the edges of the building, using theHough transform to get the straight line segment of the building edge, combining the brokensmall segment that supposed to be one straight line, adjusting the dip angle of segment byright angle constraint, then the building edge information that conforming to the actualsituation was gained.Based on airborn LIDAR point, without the other supporting data, it adopts ArcGIS、ERDAS to process digital image, uses MATLAB to conduct algorithm optimization, finallyachieved the building edge detection which is based on air LIDAR point cloud data.