A Trainable Method of Parametric Shape Description

Abstract

We have developed a trainable method of shape representation which can automatically capture the invariant properties of a class of shapes and provide a compact parametric description of variability. We have applied the method to a family of flexible ribbons (worms) and to heart shapes in echocardiograms. We show that in both cases a natural parameterisation of shape results.

Active blobs: region-based, deformable appearance models

“…Furthermore, it was difficult to use snakes for recognition because of differences in sampling and parameterization in comparing recovered descriptions. This led to the development of algorithms that enforce a priori constraints on the types of allowable deformations for motion tracking [1,8,49], deformable templates [25,55,65], trainable snakes [3,12], and deformable prototypes [46]. Such approaches provide a low-dimensional characterization of shape that enables recognition and comparison of nonrigid motions.…”

“…Rather than modeling the system as an elastic material, we can instead assume nothing about it, collect data samples of the displacements of each node, and then perform a principal components analysis (PCA) [12,35,38]. The principal directions are defined in terms of the eigenvectors and eigenvalues of the displacement covari-ance matrix.…”

Deformable templates for tracking and analysis of intravascular ultrasound sequences

“…We address this question by using Principal Component Analysis (PCA) [8]. This method has been successfully applied to design deformable models [3], [4], [2] to learn and matching image models [21] and sequences [16], [22] and, finally, to represent primitive shapes [32]. …”

Face Alignment Using Boosting and Evolutionary Search

“…Among the various technologies of face alignment, Active Shape Models (ASM) [1] and Active Appearance Models (AAM) [2] have gradually taken the stage center. ASM utilized the local texture information in search of a better template, and AAM constructed appearance models according to shape parameters and global texture constraints.…”