Extensions of the Zwart-Powell box spline for volumetric data reconstruction on the cartesian lattice

IEEE Trans Vis Comput Graph. 2006 Sep-Oct;12(5):1337-44. doi: 10.1109/tvcg.2006.141.

Abstract

In this article we propose a box spline and its variants for reconstructing volumetric data sampled on the Cartesian lattice. In particular we present a tri-variate box spline reconstruction kernel that is superior to tensor product reconstruction schemes in terms of recovering the proper Cartesian spectrum of the underlying function. This box spline produces a C2 reconstruction that can be considered as a three dimensional extension of the well known Zwart-Powell element in 2D. While its smoothness and approximation power are equivalent to those of the tri-cubic B-spline, we illustrate the superiority of this reconstruction on functions sampled on the Cartesian lattice and contrast it to tensor product B-splines. Our construction is validated through a Fourier domain analysis of the reconstruction behavior of this box spline. Moreover, we present a stable method for evaluation of this box spline by means of a decomposition. Through a convolution, this decomposition reduces the problem to evaluation of a four directional box spline that we previously published in its explicit closed form.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Computer Graphics*
  • Computing Methodologies*
  • Database Management Systems*
  • Databases, Factual*
  • Image Interpretation, Computer-Assisted / methods*
  • Imaging, Three-Dimensional / methods*
  • Information Storage and Retrieval / methods*
  • Numerical Analysis, Computer-Assisted
  • User-Computer Interface