A method of boundary estimation from 3D scattered point data without normals by implicit function and Delaunay tetrahedralization

Abstract  

A method of boundary estimation from 3D scattered point data has been proposed. For estimating a boundary, the implicit function-based method and the Delaunay tetrahedralization are mainly used in the proposed method. An advantage of the proposed method is that point coordinates are only required as input data. Namely, normals on each of given points are not required as input data. Instead, the point normals are estimated three times. After each procedure for estimating point normals, the accuracy of the estimated normals may be better. Therefore, the geometric structure of a surface generated with the estimated normals is gradually closer to the original surface. Numerical experiments demonstrate that the proposed method enables to estimate an expected boundary without normals as input data. In addition, the performance of the proposed method is numerically investigated. The estimated boundary can be obtained as an implicit surface or as a set of triangles.