Feature Selection of 3D Volume Data through Multi-Dimensional Transfer Functions

Pattern Recognit Lett. 2007 Feb 1;28(3):367-374. doi: 10.1016/j.patrec.2006.04.008.

Abstract

Direct volume rendering maps data values to visual properties such as transparency and color through transfer functions. Traditional multi-dimensional functions are generated based on a 2D histogram of function value and gradient magnitude. When two different features overlap in the 2D histogram, the traditional transfer functions cannot visually distinguish the features, since overlapped areas have similar visual properties. In this paper, we describe a new multi-dimensional transfer function that enables visual differentiation of features even in the case when two different features overlap in the 2D histogram. Furthermore, we provide details of an implementation of our transfer function on modern programmable graphics hardware.