Abstract

In this paper the implementation of a new feature-preserving volume-filtering technique is presented. The method is based on the minimization of a three-component global error function penalizing the gradient and density deviations and the curvature of the unknown filtered function. This method performs filtering in the frequency domain. Therefore, an effective 3D Fourier transformation was necessary to be implemented. We describe briefly the basics of the Fourier transformation and its optimizations. The filtering method was implemented in Matlab, Java and C. For the sake of clarity, we explain the reason why we have implemented these three versions, as well as the problems that have arisen during the implementation. We will describe the functionality of the analyzing tool implemented in Java and the optimized algorithm in C. At the end we shortly sketch some other application possibilities.

KEYWORDS: image processing, fast Fourier transformation, antialiasing, noise filtering, feature-preserving smoothing.