Research On Parametric Modeling Method Of Point Cloud Based On The Data Driven Approach

With the development of cloud computing devices,such as airborne lidar,ground borne lidar,backpack and handheld scanners and other hardware devices,point cloud data has become an important data source of geographic information.And it is widely used in digital city,resource exploration,topographic mapping and so on.At present,the use of point cloud data for building reconstruction is particularly common work.The 3D reconstruction of point cloud data is mostly based on data driven approach,which is based on the extraction of refined building point cloud and the formation of 3D model by point cloud.Due to the complexity of the building,parametric modeling has its limitations.However,parametric modeling is accurate and easy to store.So it is necessary to study the parametric modeling of point cloud data.This paper will set the building reconstruction as goal and focus on the research of point cloud segmentation and parametric modeling.The main research contents of this paper are as follow:1.Introduces the pretreatment of building point cloud.And mainly introduces the organization of point cloud data,and the basic point cloud filtering.At the same time,the slope information entropy is used as the main criterion to extract the building point cloud.2.Introduces the dimension information of point cloud to solve the problem that the traditional region growing plane segmentation method may have,and presents an improved region growing algorithm.3.Uses ?-shape algorithm to extract the edge points of buildings point cloud.Using the sleeve algorithm to simplify the edge points and get the outline information of the building.And design the parameters of the outline of the building.4.Extract the ridge lines by the method of plane intersection.And design procedures to detect the small structure of the point cloud according to the characteristics of the small structure.And then The small structure is treated separately.5.Design a set of parameters for the representation of the whole building and then reconstruct the buildings by these parameters.6.Makes some experiments to verify the effectiveness of the proposed method.And the results are compared with the terraSolid software results.