Modeling nonrigid moving 3-D objects entails recovery of global motion, global shape, and local shape. The already difficult problem of tracking nonrigid local shape is made worse by the need for a model-data correspondence scheme to make tracking possible. Both problems are addressed. Various methods for establishing 3-D model-data correspondence are discussed, and a correspondence scheme for tessellated data is presented. A force-based method for nonrigid motion tracking is presented. The authors' method does not use complex finite element techniques but follows a residual force-based dynamic model for local deformation. Another novel feature is that the net forces are obtained by an adaptive force distribution scheme which ensures smooth and stable deformation. Results are presented of tracking simulated as well as real 3-D data of the inner wall of the heart ventricle from segmented magnetic resonance images.<>
3-D model-data correspondence and nonrigid deformation
01.01.1993
176405 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Point correspondence in unstructured nonrigid motion
IEEE | 1995
|Point Correspondence in Unstructured Nonrigid Motion
British Library Conference Proceedings | 1995
|Curvature-Based Approach to Point Correspondence Recovery in Conformal Nonrigid Motion
British Library Online Contents | 1994
|British Library Online Contents | 2003
|Model learning and recognition of nonrigid objects
IEEE | 1989
|