Research On Scattered Point Cloud Registration And Implicit Surface Reconstruction Technology

In recent years,reverse engineering technology has been widely applied in the field of modern design and manufacturing,virtual reality,scenario modeling,computer vision,medicine and so on,it has very broad prospects.Point cloud processing and surface reconstruction is a key content in reverse engineering.With the wide application of three-dimensional laser scanner,how to process large-scale and densely distributed point cloud and how to reconstruct the surface with scattered point cloud has become a hot issue of the current research.The research contents of this thesis are mainly divided into two aspects:point cloud registration and point cloud surface reconstruction.The main contents are as follows:(1)In point cloud registration,this thesis proposed an new algorithm of point cloud for dimension-reduced,which aimed to solve the problem of low efficiency and noise affecting the algorithm greatly.In this method,the total projection entropy,a new method for evaluating the spatial location of point clouds,is used to evaluate the relative position of two point clouds by combining the concept of information entropy with the spatial coordinates and density relations of point clouds.The total projection entropy is used as the optimization function of point cloud registration,and the genetic algorithm is used as the optimization method.The point cloud registration is realized by combining the two methods.Experiments show that this method has strong robustness,and has good effect on only a few overlapping point cloud data and noise point cloud data.This method is efficient and can provide excellent original point cloud data for surface reconstruction of point cloud.(2)In the aspect of point cloud surface reconstruction,aiming at the problem that the previous scattered point cloud surface reconstruction algorithms are difficult to deal with large-scale point cloud data,excessive human participation and poor effect on non-closed model reconstruction,an implicit surface reconstruction method for three-dimensional point cloud data is proposed based on adaptive octree and improved differential evolution algorithm.Firstly,adaptive octree is used to decompose and process point clouds with tens of thousands of points by providing point clouds data of partitioned regions related to model density.Secondly,the improved radial primitive sphere model is used to establish the local implicit surface function to ensure the smoothness of the local surface,and the differential evolution algorithm is used to solve the radial basis center,influence radius and shape parameters adaptively.Finally,the improved logarithmic exponential weighted stitching algorithm is used to smooth the local surface,and the moving cube algorithm is used to draw the complete implicit surface.Experiments show that this method not only has good adaptability to many kinds of point clouds,but also can reconstruct the surface with smooth and obvious details.