A Hierarchical Deformable Model Using Statistical and Geometric Information

A new deformable model has been proposed by employing a hierarchy of affine transformations and an adaptive-focus statistical model. An attribute vector is used to characterize the geometric structure in the vicinity of each point of the model; the deformable model then deforms in a way that seeks regions with the similar attribute vectors. This is in contrast to most active contour models, which deform to nearby edges without considering the geometric structure of the boundary around an edge point. Furthermore, a deformation mechanism that is robust to local minima is proposed, which is based on evaluating the snake energy function on segments of the snake at a time, instead of individual points. Various experimental results show the effectiveness of the proposed methodology.