In traditional shape-based deformable models, the external image forces come primarily from edge or gradient information. Such reliance on edge information, however, makes the models prone to get stuck in local minima due to image noise and various other image artifacts. Integrating region statistics constraints has been a centerpiece of the efforts toward more robust, well-behaved deformable models in boundary extraction and segmentation. In this chapter, we review previous work on the loose coupling of boundary and region information in two major classes of deformable models: the parametric models and the geometric models. Then, we propose a new class of deformable shape and texture models, which we term Metamorphs. The novel formulation of the Metamorph models tightly couples shape and interior texture and the dynamics of the models are derived in a unified manner from both boundary and region information in a variational framework. © 2006 Springer Science+Business Media, Inc.
CITATION STYLE
Metaxas, D., Huang, X., & Chen, T. (2006). Integrating shape and texture in deformabie models: From hybrid methods to metamorphs. In Handbook of Mathematical Models in Computer Vision (pp. 113–129). Springer US. https://doi.org/10.1007/0-387-28831-7_7
Mendeley helps you to discover research relevant for your work.