Abstract
Physiologically correct animation of the heart should incorporate nonβhomogeneous and nonlinear motions of the heart. Therefore, we introduce a methodology that estimates deformations from volume images and utilizes them for animation. Since volume images are acquired at regular slicing intervals, they miss information between slices and recover deformation on the slices. Therefore, the estimated finite element models (FEMs) result in coarse meshes with chunk elements the sizes of which depend on the slice intervals. Thus, we introduce a method of generating a detailed model using implicit surfaces and transferring a deformation from a FEM to implicit surfaces. An implicit surface heart model is reconstructed using contour data points and then crossβparameterized to the heart FEM, the timeβvarying deformation of which has been estimated by tracking the insights of the heart wall. The implicit surface heart models are composed of four heart walls that are blended into one model. A correspondence map between the source and the target meshes is made using the template fitting method. Deformation coupling transfers the deformation of a coarse heart FEM model to a detailed implicit model by factorizing linear equations. We demonstrate the system and show the resulting deformation of an implicit heart model. Copyright Β© 2008 John Wiley & Sons, Ltd.