Deformable Models for Object Matching and Tracking

We   propose   a   general   object   localization   and   retrieval   scheme   based   on   object   shape   using  deformable templates. Prior knowledge of an object shape is described by a prototype template which  consists of the representative contour/edges, and a set of probabilistic deformation transformations on the template. A Bayesian scheme, which is based on this prior knowledge and the edge information in the  input image,  is employed to find a match between the deformed template  and objects in the  image.   Computational  efficiency is  achieved  via  a  coarse-to-fine   implementation  of   the  matching  algorithm. Our method has been applied to retrieve objects with a variety of shapes from images with complex  background.  The proposed scheme  is  invariant to location, rotation,  and moderate scale changes of the template.