Shape Modelling Using Markov Random Field Restoration of Point Correspondences

Rasmus R. Paulsen, Klaus B. Hilger

AbstractA method for building statistical point distribution models is proposed. The novelty in this paper is the adaption of Markov random field regularization of the correspondence field over the set of shapes. The new approach leads to a generative model that produces highly homogeneous polygonized shapes and improves the capability of reconstruction of the training data. Furthermore, the method leads to an overall reduction in the total variance of the point distribution model. Thus, it finds correspondence between semilandmarks that are highly correlated in the shape tangent space. The method is demonstrated on a set of human ear canals extracted from 3D-laser scans.
KeywordsMarkov Random Field, Statistical Shape Model, Registration, Thin Plate Spline
TypeConference paper [With referee]
ConferenceInformation Processing in Medical Imaging
Electronic version(s)[pdf]
BibTeX data [bibtex]
IMM Group(s)Image Analysis & Computer Graphics