Biomedical Engineering Reference
In-Depth Information
Chapter 10
A Region-Aided Color Geometric Snake 1
Xianghua Xie 2 and Majid Mirmehdi 2
10.1 Introduction
Deformable contour models or snakes are commonly used in image process-
ing and computer vision due to their natural handling of shape variation and
independence of operation (once initialized). A hypothesized contour, repre-
sented as a curve or surface, evolves under the influence of internal forces,
external image-dependent forces, and certain constraints, till it converges on
the object(s) of interest.
Generally, there are two types of snakes, parametric snakes and geometric
snakes. The parametric model minimizes a deforming curve toward the pull of
features such as edges and lines. The energy is composed of terms that con-
trol its smoothness and attract it to the object boundary. Although significant
improvements have been made in this field over the last decade, parametric
contours still suffer from imprecise shape representation. The geometric model
of active contours, which avoids the need to parameterize the curve, has been
hailed as the solution to topological problems. Geometric snakes are based on
the theory of curve evolution and are numerically implemented via the level set
algorithm. They are totally intrinsic, which means they can automatically han-
dle topological changes without resorting to dedicated contour tracking, and
unknown numbers of objects can be detected simultaneously. Furthermore,
they can enjoy much larger capture areas than parametric snakes.
1 Portions reprinted, with permission, from IEEE T-IP, 13(5): 640-652 by the same authors.
2 Xianghua Xie and Majid Mirmehdi Department of Computer Science, University of Bristol,
England.
535
 
Search WWH ::




Custom Search