The Method Study Of Building Extraction Based On Airborne LIDAR Point Cloud Data

Airborne LiDAR,as a new type of remote sensing sensor,it can quickly and directly obtain three-dimensional coordinate information.Meanwhile,with the acceleration of the process of Digital City,it’s gradually become a new trend of study on urban buildings based on fastly and accurately data source.Building as one of the important characteristics of urban surface area,it has lots of important applications on modern urban planning and construction,disaster prevention,three-dimensional digital simulation,military reconnaissance and many other areas.This paper aims to extract the building area with airborne LiDAR point cloud data after generating DEM depth image,for the feature of experimental data area in this paper of the main buildings mainly to low houses and many trees close to the building area or together with the buildings,this paper presents a method that combine the region growing segmentation with mathematical morphology for extracting the building,the main contents and conclusions include:(1)Based on the analysis on each of the interpolation method comparative tests using a weighted average interpolation for the test area’s date with different grid interpolation size,comparative analysis in two aspects on the treatment time and the effect of the interpolation,interpolation processing time is reduced with the decrease of the interpolation grid size and when the interpolation grid size is 0.1m the interpolating effect of this paper data is best.(2)Through comparing smoothing filter tests,we can derive that median filtering can achieve the purpose that protecting the building edge information is not destroyed and removing some part of non-building area at the same time.(3)This paper presents a method that combine the region growing segmentation with mathematical morphology for extracting the low building,and improve on the basis of mathematical morphology processing method,propose a segmentation method based on region growing and add to remove small areas in the open operation to exclude non-building area step by step,especially those non-building area with the height higher than the buildings and larger area and close to the building edge,and make up for the void phenomenon between buildings,get the ideal building region.(4)We study seven common edge detection operators,and on the basis of analysing the insignificance of the traditional Canny operator.Then the traditional Canny operator has been improved.Through the simulation experiment and comparative experiment proving that the improved Canny operator can obtain better positioning accuracy and better than the other seven kinds of edge detection algorithm on noise immunity.Through the comparition of building edge of experimental extraction and depth superimposed image,conducted accuracy assessment from two indicators of accuracy and correctness.