Three-dimensional image analysis and display by space-scale matching of cross sections

Abstract

We present a framework and a set of techniques for the analysis and display of three-dimensional experimental data or images. We assume that the data are available in the form of two-dimensional cross sections of the three-dimensional data set. We describe our approach, which has for goals to extract significant information from the three-dimensional data set and to display this information as objects that can be manipulated in three-dimensional space. The high-contrast transitions of two-dimensional cross sections are extracted first; they define a set of contours to be matched from cross section to cross section. This matching is performed by space-scale analysis of the orientation of contours on adjacent cross sections. By modeling the contours as B splines, we then make use of three-dimensional B-spline patches to generate significant surfaces that can be displayed, rendered, and rotated with standard computer graphics techniques and specialized processors.