PolyHist: A polyhedral histograming technique for volumetric data analysis

Graduate

Author: Paul Albee
Advisor: Dr. Stockman
Email: albeepau@cse.msu.edu; http://www.cse.msu.edu/~albeepau

Micro-tomographic analysis of soil aggregates has emerged as a promising area in soil science. One area of interest to soil scientists is to identify well defined density changes in the soil aggregates. Non-uniform density implies the presence of pores in the aggregates that allow the transport of material through the aggregate. A technique, PolyHist, is described which aids in determining if non-uniformities are present in a volumetric data set. PolyHist applies a set of feature operators to the data, then maps the results onto a convex polyhedron. The resulting map can be used to determine how uniform or interesting the volume is with respect to the applied feature set. This technique is well suited for parallel processing since the results are easily combined, and only local information is needed for computing each sub-volume. Additionally, this technique does not require the computation of iso-surfaces to identify the presence of density changes, allowing for faster analysis. We present the results of applying our technique to several soil aggregates data sets collected at Argonne National Laboratory.

Return to Workshop Listing