Information Technology Reference
In-Depth Information
5
10
1
2
3
4
9
6
13
7
11
12
8
Figure 11.2
Subimages for Semi_Global histogram. (From ISO/IEC TR 15938-8:2002 Information
t e c h nolo g y.)
This matching method is a good example of the distance calculation of
the derived secondary features from the primary feature available in a
descriptor.
11.2.3 Shape Feature Descriptors
Shape is a prominent feature of image objects perceived by humans and pro-
vides powerful visual characters of images for similarity matching.
The current application of these feature descriptors includes binary
images, trademarks, and presegmented images. The application of shape fea-
ture descriptors is limited by the availability of shape information in images,
which is hard to obtain automatically from ubiquitous images and videos.
The more detailed description of these descriptors can be found in ISO/IEC
information technology [1].
The MPEG-7 standard defines several feature descriptors based on shape.
•
Region Shape Descriptor
The region shape descriptor specifies any type of shape ranges of
an object in images from a simple shape like a triangle or circle
to complex ones like a donut or a trademark. It uses a set of ART
(Angular Radial Transform) coefficients and robust to noises that
may be introduced in the process of segmentation. It allows minor
nonrigid deformations, split objects, and even an object with sepa-
rated regions.
•
Contour Shape Descriptor
The contour shape descriptor specifies a closed contour of a 2-D
object or region in images. It is based on the Curvature Scale Space
(CSS) representation of the contour extracted from a list of contour
points of an object or region. It can distinguish between shapes that
have similar region-shape properties but different contour-shape
properties and robust to noise, nonrigid deformations, and distor-
tions in the contour due to perspective transformations.
