A Comparison between Moving Least-Square Method and Natural Neighbour Interpolation

Article Preview

Abstract:

The construction procedures of shape functions in the moving least-square method (MLS) are complicated, in which many matrix multiplication and matrix inversion are included, so that the computational efficiency is low. Moreover, the choices of some parameters are influenced by the artificial factors, and the computational stability is poor. However, the construction procedures of shape functions in natural neighbour interpolation (NNI) are based on Voronoi diagram and its dual Delaunay triangulation, computational results are only related with the locations of the discretized nodes, and the computational stability is good. In order to study the differences in the computational accuracy, the computational efficiency, and the adaptability to the fitted objects between MLS-based shape functions and C1 natural neighbour interpolant, the two higher-order continuous shape functions are introduced in surface fitting.

You might also be interested in these eBooks

Info:

Periodical:

Pages:

653-656

Citation:

Online since:

August 2013

Export:

Price:

[1] P. Lancaster and K. Salkauskas: Math. Comput. Vol. 37 (1981), p.141.

Google Scholar

[2] R. Sibson: Math. Proc. Cambridge Philo. Soc. Vol. 87 (1980), p.151.

Google Scholar

[3] G. Farin: Comput. Aided Geom. Des. Vol. 7 (1990), p.281.

Google Scholar

[4] N. Sukumar and B. Moran: Numer. Meth. Part. Diff. Eq. Vol. 15 (1999), p.417.

Google Scholar

[5] Z. F. Nie, S. J. Zhou, R. J. Han, L. J. Xiao and K. Wang: Acta Mech. Sin. Vol. 28 (2012), p.91.

Google Scholar