Abstract:With the development of laser scanning measurement technology,the detailed information about the surface point cloud data of the geometric model is more abundant due to the more efficient data detection accuracy,make it more precise to show the surface features of objects.However,the corresponding technical challenges may appear at the same time because of such a large amount of point cloud data,which can be used in the computer file storage,data post-processing and software visualization inconveniently and inefficiently.A new algorithm is introduced in this paper.Firstly,we make a space division for point cloud data and establish the domain relationship using the grid method.Secondly,we estimate the point cloud normal vector by means of local surface fitting.Thirdly,we find out the significant value of the coordinate points using the point cloud K field method.Finally,we achieve the point cloud octree according to the significant value.In a word,this algorithm realizes the goal that the significant features of the point cloud can be extracted and the amount of the point cloud data can be simplified.Not only does it retain the advantages of the detail characteristics of the point cloud,but also make it more effective.