Study On The Building Extraction From Airborn LiBAR Points Cloud Data

LiDAR(Light DetectAnd Ranging) as a new type of remote sensorscan rapidly acquirean accurate three-dimensional coordinate information and digital surface model(DSM).So thetechnology of LiDARenhances the accuracy and speed of survey and automate acquisitionandprocessing of the data.Meanwhile, with the accelerated pace of digitization town, the studyof the extraction of city buildings via the fast and accurate data sources has become a newtrend. It also becomes an important reason for promoting the LiDAR data using in urban areas,especially in the city building. Building is one of the major urban surface features，and it hasimportant applications in some area,such as modern urban planning and construction,disasterprevention,exploration and utilization of natural resources,digital city construction andthree-dimensional numerical simulations and military reconnaissance. Therefore, the study ofhow fast and accurate automatic extract the buildinghas a very important practicalsignificance.In this paper, with the LiDAR point cloud data as the object of study,on the basis of thecharacteristics of point cloud data in brief,studies the irregular discrete LiDAR point clouddata,which result in digital surface model (DSM).Then,use the image processing techniques topreprocess DSM image. After that, make a contrast test for the preprocessed DSM image inresult of threshold segmentation and clustering segmentation, then choose the bettersegmentation result as the image for subsequent processing. Finally, use theedgedetectionoperatortor toextract the DSM depth segmentation images. Using the Houghtransform to merge the broken small edge of the building finds edge information.Finally,compared with the DSM depth images,analysis its accuracy and factors affecting the accuracyof edge detection.In short,based on air LiDAR point, without the other supporting data, it adopts ArcGIS、ERDAS、 MAPGIS to process digital image, uses MATLAB to simulate algorithmoptimization, with three sets of comparative analysis, hoping to find a simple edge detectionprocess based on building aviation LiDAR point cloud data, and to reach a certain accuracy.