“Eccentric radial basis functions and its applications” by Kanamori, Takaoki and Nishita
Conference:
Type(s):
Title:
- Eccentric radial basis functions and its applications
Presenter(s)/Author(s):
Abstract:
The use of radial basis functions (RBFs) has a long history in computer graphics. Typical applications of RBFs are continuous interpolation/approximation of discrete data, and field functions of metaball-like implicit surfaces. In this paper we introduce eccentric radial basis functions (ERBFs) in order to enhance the power of expression of RBFs. An ERBF is an RBF equipped with a point called ”eccentric center”, which is used to compute the distance for the RBF and introduce eccentricity into the distribution of the function values; the gradient of the function becomes gentle at the far side of the eccentric center while steep at the near side of it. The extension is simple yet powerful, and allows us to fit data with fewer basis functions than RBFs, especially when fitting or modeling data consisting of both gentle and steep gradations. We demonstrate the effectiveness in applications such as fitting of scattered data and modeling with ERBF-based metaballs.
References:
1. Bloomenthal, J., and Wyvill, B., Eds. 1997. Introduction to Implicit Surfaces. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA.
