Research On Laser Point Cloud Data Compression And Surface Modeling

It is one of the main methods to generate the 3D digital model based on various data acquisition equipment in recent years,and has been widely used in surveying,computer vision,computer graphics,archaeology and other fields.The development of modern science and technology makes the 3D laser scanning hardware equipment developed rapidly,however the development of corresponding data processing technology is relatively slow,has become a bottleneck and restricted the further development of 3D laser scanning technology,especially for the point cloud data storage,processing,transmission and further application.Therefore,the development of 3D laser point cloud data processing theory and methods has become the focus of academic research.In this paper,the 3D laser scanning technology and the method of data acquisition and processing are introduced in detail,and the algorithm of point cloud data compression and surface modeling is deeply studied,the main research work is summarized as follows:Firstly,aiming at the 3D laser scanning equipment often produces high density and information redundancy of point cloud data.However,the existed algorithms are insufficient.Extending the point cloud data point by point compression of one dimensional scanning line in the coordinate increment method to two-dimensional scanning lines between.Thus point cloud data compression algorithm based on improved coordinate increment has been proposed on the basis of studying the existing algorithms.A case study is conducted in Matlab to compare the compression effect of the proposed method and several existing typical compression method such as coordinate increment algorithm,curvature sampling algorithm,barycenter of area data compressing method,random sampling algorithm and so on.The experiments show that the proposed method achieves good compression effect for point cloud data of scan-lined plane or curved surface.Secondly,combined the method of qualitative and quantitative to analyze the existing algorithms and improved compression algorithm results from precision,simplicity,speed and generality,verified the feasibility of improving the compression algorithm.Then based on Geo magic Studio software to modeling the point cloud data before and after compress,analyzed the compression effect of this method from the change rate of surface area and 3D standard deviation,the fidelity and reliability of this algorithm are further verified..Finally,aim at surface modeling,mainly studies the surface fitting methods of typical region growth method,B spline surface fitting algorithm and the back-propagation neural network method,and summarized the limitations of the existing surface fitting method,then based on Cyclone and Geomagic Studio software reconstructing the 3D model of the big wild goose pagoda and the Lugou Bridge stone lion,and the deficiency in the process of reconstruction is analyzed.Thus providing a reference for the storage,management,analysis and display of massive point cloud data in the era of big data.