A geometrical approach for surface reconstruction by means of radial basis functions with compact support

Abstract

Recently in the community of computer vision related with the surface reconstruction processes of free-form objects there has been a growing trend in the use of interpolation techniques. In this area the radial basis functions interpolator can produce three-dimensional models with high levels of precision, high flexibility to reproduce complex shapes and a high tolerance to noise level. The radial basis functions used for data interpolation need to estimate a set of parameters. This work shows a approach to estimation of those parameters based on the surface geometric characteristics, these are both centers and support ratio. The approach results are shown mean by interpolation of real objects range data.