Research On Point Cloud Classification Algorithm Based On Gaussian Mapping

With the concept of "digitalization," in recent years,two-dimensional and three-dimensional,real and virtual environmental integration capabilities have been increasingly valued by people.Because of its high data sampling rate,non-contact measurement,strong anti-interference,and high precision,three-dimensional laser scanning has been gradually applied in various fields.Nowadays,the research on 3D laser scanning technology is becoming more and more mature,and the processing technology of point cloud data needs to be improved.In the preprocessing of point cloud registration,point cloud denoising,and point cloud streamlining,there are still poor processing results.Problems such as low processing efficiency.With the continuous improvement of the accuracy of point cloud data processing and processing efficiency in all walks of life,a deeper level of research is needed on the point cloud data processing technology.The point cloud data acquired by the three-dimensional laser scanner is usually a scattered point,and there is no spatial topology information.Before processing the point cloud data,the spatial topology of the point cloud data must be constructed by obtaining the local topology of the point cloud data.,The local neighborhood of point cloud data is crucial for subsequent processing.In this paper,according to the advantages and disadvantages of the existing point cloud data local neighborhood search algorithm,the search range in the local neighborhood search can not be determined and the search loop may be encountered and solved.In addition,for the point cloud classification process,different types of point cloud data need to use different classification methods.This paper introduces the Gaussian mapping into the point cloud data classification process,and strives to find a point with stronger applicability.Cloud data classification algorithm.For the issues raised above,the main work and achievements of this paper are as follows:(1)The current point cloud data local neighborhood search algorithm is studied and its advantages and disadvantages are analyzed.The specific problems of discov-ery are studied in depth and a solution is proposed.The solution is implemented using C++ language and validated by comparison.The improvement is obtained through experiments.The latter algorithm has a clear enlargement rule for the range of the neighborhood search,and avoids the occurrence of an infinite loop in the search.Compared with the existing algorithms,the efficiency of the improved algorithm has also been greatly improved.(2)The specific research on the point cloud data normal vector,the Gaussian map of point cloud data,and the DBSCAN clustering algorithm are studied.The principles are mastered and implemented.(3)Select some point cloud data of Yunxia ancient temple in Neijiang City to conduct experiments.Firstly,the point cloud data is solved with normal vectors and Gaussian mapping is performed.The DBSCAN algorithm is used to cluster the data on the Gaussian ball,and the clustering result is finally reflected.In the original point cloud data,the classification of the point cloud data was completed and a more satisfactory result was achieved.