"A Landmark-free Method for Three-Dimensional Shape Analysis"
Benjamin J. Pomidor
Department of Scientific Computing
Florida State University
The tools and techniques used in shape analysis aim to transform the physical shape of an object into a concise set of numerical data for mathematical modeling and statistical analysis. The advent of landmark-based morphometrics opened new avenues of research in this area, but these methods are not without drawbacks. The time investment required of trained individuals to accurately landmark a data set is significant, and the reliance on readily-identifiable physical features can limit research, especially when investigating smooth or featureless surfaces.
In this talk, we present a new method, based upon and extending the Iterative Closest Point algorithm, to perform this transformation for data obtained from high-resolution scanning technology. This method uses surface scans, instead of landmarks, to calculate a shape difference metric analogous to Procrustes distance and perform superimposition. We also explore some new ways this data can be used; for example, we can calculate an averaged surface directly and visualize point-wise shape information over this surface. We demonstrate the method on a set of primate skulls and compare the results of the new methodology with traditional geometric morphometric analysis.