Feature Point Extraction And Building Reconstruction Based On 3D Laser Scanning Point Cloud Data

Modeling and processing of Point cloud data is an important research content of laser scanning system.For different target object,the modeling way are not same.Commonly used way of modeling is fitted and modeling on the basis of grid building,there can achieve good effect in large quantity of surface change and complex shape objects modeling,but the construction mesh needs large amount of calculation,the modeling method is only applicable to less data objects.For buildings,large amount of data,and usually the shape is rules,Therefore,a professional building modeling software for building modeling is necessary,you can put point cloud data imported into the modeling software directly,then option manually to model building,since there are large amount of point cloud data source and the choice data process is completely artificial,the model quality is decided by the modeling staff's level,modeling precision cannot be guaranteed.However,the modeling method that based on feature points,is based on building feature to modeling,the modeling method can improve the degree of automation,make sure the accuracy of modeling,and reduce the precision loss of artificial error.This paper describes the working principle of the three-dimensional laser scanning based on the ground,analysis the reason of error produce,and discuss the method of reducing the error.preprocessing of point cloud data,including the point cloud registration,point cloud denoising,point cloud streamline and data segmentation,the different preprocessing methods are discussed in this paper,and the advantages and disadvantages of each method has discussed.Then,using Geomagic Studio software to processing the original point cloud data,such as data noise and compaction.two kinds of feature point extraction methods has been illustrated include of curvature estimation,principal component analysis,using Matlab programming for principal component analysis to feature point extraction of point cloud data,using Imageware software for curvature estimation to extract the feature points,then to perfect feature point extraction by manually modify.Then,put point cloud into 3DMax and modeling directly on the point cloud data,and then,modeling by Sktechup software that based on feature points extraction of point cloud data.Eventually,comparing the two models with original point cloud data by Geomagic Control,generating standard deviation statistics that standard distribution,the deviation value is concentrated in a certain range,which indicates that both modeling methods are feasible.It can be seen from the 3D results comparison table,for the four deviations,modeling based on the feature extraction method are smaller than the manual selection method,the experiment proved that the method based on feature extraction model can effectively reduce the error,modeling precision is better,can effectively improve the modeling efficiency.