Research On Simplification Method Of 3D Point Cloud Based On The Importance Of Point

With the rapid development of computer vision,people have higher and higher requirements for the acquisition of 3D information.3D point cloud scanning technology continues to break through,and high-precision data acquisition instruments continue to appear.The following problem is that there must be a large number of redundant points in the high-precision and high-density point cloud data set,resulting in waste of resources,increase of computing costs,and having a certain impact on the point cloud subsequent processing.Therefore,it is necessary to simplify 3D point cloud data,which can not only save computer resources,but also improve the subsequent processing speed of point cloud.According to the simplification form,the existing point cloud simplification algorithms can be divided into two types: point cloud simplification based on grid and point cloud simplification based on point.However,no matter which form of point cloud simplification algorithm,there are some defects.The simplification algorithm based on grid needs to generate a large number of grid,the computer cost is high and it costs a lot of time and space.The simplification algorithm based on point takes up less computer resources,but it is easy to lose the detail features of the original point cloud data set in the process of simplification.In addition,large-scale simplification is bound to cause the loss of point cloud features,and the high retention of features is bound to reduce the simplification rate.Therefore,in the process of point cloud simplification,how to delete as many redundant points as possible under the premise of retaining the original point cloud data features has beena hot and difficult research topic of researchers for many years.Moreover,for different point cloud data sets,the simplification algorithm should be flexible to adapt to different simplification requirements.In this paper,the simplification methods of 3D point cloud data are studied systematically.Aiming at the problem of data redundancy of 3D point cloud,the simplification method of 3D point cloud based on the importance of point is proposed,its main innovations include the following three aspects:(1)Aiming at the limitation of traditional neighborhood search method using fixed search order,the k neighborhood search method based on distance and density is proposed to ensure that the neighborhood points searched are the closest to the sample points.(2)Aiming at the problem of feature loss in the process of point cloud simplification,the feature retention method based on the importance of points is proposed.According to the geometric features of point cloud,four feature operators are given to evaluate the importance of points from multiple perspectives,and then the feature points of the whole point cloud data set are screened to retain the detailed features of the original point cloud data set.(3)Aiming at the problem of incomplete simplification of non-feature points,the method of non-feature point simplification based on octree is proposed.For the non-feature point set,the octree is constructed,and the points in each octree leaf node are highly simplified to ensure the simplification rate of point cloud.In this paper,experiments are carried out on four point cloud data sets,including two data sets with less feature information,bunny data set and horse data set,and two data sets with more feature information,gargoyle data set and elephant data set.The simplified point cloud data can retain most of the feature information of the original point clouddata,and the reconstruction error is small.In addition,this algorithm is compared with several traditional simplification algorithms,and the experimental results show that the algorithm in this paper has better simplification results,and is more suitable for simplifying point cloud data sets with a large number of features.Point cloud simplification is the research focus of this paper.In this paper,the idea of divide and conquer is used to deal with feature points and non-feature points respectively.The feature points are selected and retained through the geometric characteristics of points.At the same time,the method of uniform simplification is adopted for non-feature points.On the premise of retaining the characteristics of the original point cloud model,ensure that the simplification rate of the point cloud data set reaches a certain standard.In addition,the parameters of this algorithm can be determined according to the specific experimental requirements and point cloud types,and the algorithm has strong flexibility.