Boundary Extraction from Voxel Data Using Contour Lines

Richard Guercke, Leibniz Universität Hannover, masters thesis
11/2007

In this thesis, an algorithm is introduced that extracts a boundary surface from a voxel data set representation of a shape. To achieve this objective, a connectivity graph is constructed from boundary polygons of the object in (usually orthogonal) 2D slices. This graph is used to find facets of the object's boundary surface.

At given sampling intervals, the resulting triangulated surfaces bear a considerably closer resemblance to the original voxel data object than the result of the (modified) Marching Cubes algorithm.  This is, however, only achieved at the cost of an increase in computation time and memory usage. At the closest sampling interval, the algorithm returns a closed surface without self-intersections.

The connectivity graph can also be used for the restriction of surface patches to be displayed according to visibility or as a basis for physical computations like FEM analysis.

Project: YaDiV

Kontakt: Karl-Ingo Friese