Study On The Surface Reconstruction On Huizhou Architecture Component Point Cloud

Surface reconstruction technology is widely used in reverse engineering,machine vision,virtual reality,modern medical and other fields.With the rapid development of3 D laser scanning technology and the increasingly mature,3D point cloud data through the three-dimensional laser scanning technology to obtain high density surface of the ancient architectures,and then through the optimization of the segmentation,filtering denoising,normal computation,surface reconstruction and other operations to complete the three-dimensional modeling of point cloud data.Many problems such as complex operation,too many interactive definitions and too strict input data are needed to support the reconstruction of surface reconstruction software,so far it has not been widely promoted.However,the usual practice for three-dimensional modeling of ancient buildings is the pre-measured by artificial specific physical parameters of each component of the building,then using 3D modeling software system description,mapping,the final effect is not true and much error.Based on the size of components of Huizhou Arichitecture construction study,this topic selects the unique Huizhou Architecture Chuandou wooden beam framework,Menlou,Langan as the research objects to study.Research on surface reconstruction based on point cloud component of Huizhou Architecture mainly discusses the following three aspects:First,point cloud data preprocessing.Based on the characteristic of the Huizhou architecture Chuandou wooden beam framework as an example,the multi site scanning to obtain the original point cloud data.First of all,through improved iterative closest point algorithm for multiple sites cloud data registration and stitching to obtain the whole building point cloud data.Secondly,after the registration of point cloud data using random sampling consistency algorithm segmentation optimization and obtain the beam-column point cloud data.Finally,the method of voxel grid is used to sample the outliers in the point cloud data,and then the filter Statistical OutlierRemoval is used to eliminate the outliers.Second,surface reconstruction based on preprocessing of point cloud data.By using the least square method to estimate the point cloud data after preprocessing,we can get the normal and positive or negative direction of each point in the data.With the help of point cloud library open source platform,respectively,using a greedy projection triangulation algorithm,BallPivoting algorithm and the algorithm ofPoisson surface reconstruction of the beam-column point cloud data after pretreatment with normals and give the key steps of the algorithm,finally get the effect diagram and the corresponding parameter information for each algorithm.Third,the comparison of surface reconstruction results.By comparing the three algorithms under the condition of the same point cloud components the surface reconstruction of surface rendering and grid number,grid vertex number,the number of holes,the parameters such as time consuming information,results show that the algorithm of poisson final rendering significantly better than the other two methods,especially the final result,there is no hole in the 3 d model for surface reconstruction ultimately the integrity of the output is very important.The research with the aid of 3 d laser scanning technology to obtain the original point cloud data,object by polygon mesh technology such as segmentation of point cloud data,filter,normal estimation,such as pre-processing,eventually establish based on poisson algorithm has been gloriously enrolled Huizhou Architectural component surface reconstruction point cloud data model.In this paper,the whole research project of surface reconstruction is provided for the application of 3D modeling technology in architecture,especially in the field of ancient building reconstruction and digitalization provides a new research idea.