A Novel Approach for Speeding Up RBF-Based Interpolation Methods

A popular and effective method for interpolation of data is by means of radial basis functions, but the most significant drawback of this method, which made researchers unwilling to use it for practical purposes, is its severe computational cost. Introduction of different methods by the researchers active in this field, however, has alleviated this problem. This paper presents a new method for reducing the computational cost of radial basis function interpolation, which is intrinsically different from those in common use (such as center reduction and fast multipole method), and can actually be regarded as a divide and conquer solution. Also, some experiments are provided which demonstrate the efficiency of the proposed approach.