Abstract

In a companion paper [7] we formulate a broadly applicable class of mathematical models for describing the shapes of objects appearing in images. These models may be viewed as deformable bodies composed of an abstract elastic material. They are matched to images through the action of externally applied forces derived from image functions. Additional guiding forces may be applied interactively by the user. Three applications have served to illustrate the utility of the models: image based contour extraction, correspondence matching, and 3D object reconstruction. In the present paper, we deal with the discrete solution of the dynamical equations inherent to these applications. We focus on the direct and iterative numerical methods underlying our implemented algorithms for matching deformable models to images.

© 1987 Optical Society of America