Research On Hole Filling And Reconstruction Of Scattered Point Cloud

With the expansion of the application of 3D measurement technology,point cloud data acquisition,processing and reconstruction have become a research hotspot.Because of the limitation of the measurement method and the complexity of the shape of the object,holes can be shown in the point cloud.At the same time,with the improvement of the precision of the 3D measurement equipment,the magnitude of the point cloud data become larger and larger,and the efficiency of triangulation algorithm is more and more demanded by the application scene.However the existing algorithms are hard to meet the requirements.In order to solve these problems,the key technologies of hole filling and triangulation are deeply studied in this paper.And new algorithms are proposed respectively.The main work done in this paper is as follows:1.Hole repairing of scattered point cloud.Aiming at the complex holes filling problem in point cloud,an algorithm of hole repair of point cloud based on SFM is proposed.The algorithm uses the data set obtained by SFM to supplement the information of the hole area,so as to assist the hole repairing.In this method,a two-dimensional phase of fringe projection is used to extract boundary points from the three-dimensional point cloud.Then register data set obtained from SFM and fringe projection and extract supplementary point.Finally,by using the radial basis function to calculate the surface equation,the repair of point cloud holes is implemented on the point cloud data set with supplementary points.Experimental results show that the algorithm is robust and can recover the surface information of complex objects more effectively.2.Reconstruction of scattered point clouds.To improve the efficiency of mesh reconstruction of massive point cloud,a Delaunay triangulation algorithm based on improved kdtree is proposed.The algorithm improves the stopping condition of the kdtree segmentation and change the way of acquiring control points in the process of triangulation,effectively improves the efficiency of mesh reconstruction.The algorithm firstly divides the point cloud with the width first kdtree,stops the segmentation at the appropriate depth to ensure the size of the data block.And mark the points with the spatial structure.Then extract the point of the non leaf node in the kdtree structure as the control point.Finally insert the point cloud according to sequence of the control point and other points to the Delaunay triangulation algorithm to reconstruct grid.Experimental results show that the algorithm can effectively improve the efficiency of point cloud mesh reconstruction,and the quality of reconstructed mesh is better.