Scribd
Upload
43 views25 pages

Descriptors

The document discusses various descriptors that can be used to describe shapes including simple descriptors like area and perimeter as well as more complex descriptors related to topology, convex hulls, extremal points, moments, and boundary properties.

Uploaded by

Tanya Duggal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
43 views25 pages

Descriptors

The document discusses various descriptors that can be used to describe shapes including simple descriptors like area and perimeter as well as more complex descriptors related to topology, convex hulls, extremal points, moments, and boundary properties.

Uploaded by

Tanya Duggal
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 25

Descriptors

What Are Descriptors ?


• In general, descriptors are some set of
numbers that are produced to describe a given
shape. The shape may not be entirely
reconstruct-able from the descriptors, but the
descriptors for different shapes should be
different enough that the shapes can be
discriminated.
What qualifies as a good descriptor?

• In general, the better the descriptor is, the


greater the difference in the descriptors of
significantly different shapes and the lesser the
difference for similar shapes.

• So the descriptors attempts to quantify shape


in ways that agree with human intuition (or
task-specific requirements).
• Regions can either describe boundary-based
properties of an object or they can describe
region-based properties.
• In this lecture, we focus on region-based
descriptors.
Some Simple Shape Descriptors
• Area. The number of pixels in the shape. Your text
describes algorithms for calculating the area from
quad tree or chain-coding representations.
• Perimeter. The number of pixels in the boundary
of the shape.
• (Non-)Compactness or (Non-)Circularity. How
closely-packed the shape is (not): perimeter2/area.
The most compact shape is a circle (4π). All other
shapes have a compactness larger than 4π.
• Eccentricity. The ratio of the length of the
longest chord of the shape to the longest chord
perpendicular to it. (the ratio of major axis to
the minor axis ).
• Elongation. The ratio of the height and width
of a rotated minimal bounding box. In other
words, rotate a rectangle so that it is the
smallest rectangle in which the shape fits. Then
compare its height to its width.
• Rectangularity. How rectangular a shape is
(how much it fills its minimal bounding box):
area of object/area of bounding box. This
value has a value of 1 for a rectangle and can,
in the limit, approach 0 (picture a thin X).
• Orientation. The overall direction of the
shape.
Topology

Topology is the study of properties of a figure


that are unaffected by any deformation, as long
as there is no tearing or joining of the figure.
Topological Descriptors
• If we stretch a shape like we would a cutout from
a rubber sheet, there are certain shapes we can
make and others we can’t.
• Topology refers to properties of the shape that
don’t change, so long as you aren’t allowed to
tear or join parts of the shape.
• A useful topological descriptor is the Euler
number E: the number of connected components
C minus the number of holes H:
E=C-H
Convex Hull: Bays
• Although not strictly a topological property, we can
also describe shape properties by measuring the
number or size of concavities in the shape. We can
do this first finding the convex hull of the shape and
then subtracting the shape itself. What are left are
either holes (“lakes”) or concavities (“bays”). This
might be useful in trying to distinguish the letter
“O” from “C”, etc. One can then recursively identify
concavities within each “bay”, etc. The resulting
structure is a concavity tree.
Extremal Points
• Another way to describe a shape is to find its
extremal points.
• The simplest form of extremal representation is
the bounding box: the smallest rectangle that
completely contains the object.
• A more powerful way of using extremal points
is to find the eight points defined by: top left,
top right, left top, left bottom, bottom right,
bottom left, right top, and right bottom.
• By connecting opposing pairs of extremal points
(top left to bottom right, etc.) we can create four
axis that describe the shape. These axis can
themselves be used as descriptors, or we can us
• For example, the longest axis and its opposite
(though not necessarily orthogonal axis) can be
labeled the major and minor axes. The ratio of
these can be used to define an aspect ratio or
eccentricity for the object. The direction of the
major axis can be used as an (approximate)
orientation for the object.
Moments
• Another way to describe shape uses statistical
properties called moments.
• What Statistical Moments Are
• Statistical moments like mean, variance and
higher order moments can quantitatively
describe the shape of boundary segments.
BOUNDARY DESCRIPTORS
• Length of a Contour:
By counting the number of pixels along the
Contour. For a chain coded curve with unit
spacing in both directions, the number of
vertical and horizontal components plus 2 1/2
times the number of components give the
exact length of curve.
BOUNDARY DESCRIPTORS…….cont’d

• Boundary Diameter:
It is defines as
Diam (B) = maxi,j [D(pi, pj)]
where D is the distance measure which can be
either Euclidean distance or D4 distance.

• The value of the diameter and the orientation of the


major axis of the boundary are two useful
Descriptors.
BOUNDARY DESCRIPTORS…….cont’d

• Curvature:
It is the rate of change of slope.

• Curvature can be determined by using the


difference between the slopes of adjacent
boundary segments at the point of intersection
of the segments.
BOUNDARY DESCRIPTORS…….cont’d

• Shape Numbers:
• Shape number is the smallest magnitude of the
first difference of a chain code representation.
• The order of a shape number is defined as the
number of digits in its representation.
• Shape order is even for a closed boundary.
• The length of a boundary is one of the boundary
descriptors.
a) True
b) False

• Answer: a
Explanation: The length of a boundary is one of the
simple boundary descriptor. The length of the
boundary is approximately given by the number of
pixels along that boundary.
• Which of the following is the useful descriptor of a
boundary, whose value is given by the ratio of length of
the major axis to the minor axis?
a) Radius
b) Perimeter
c) Area
d) Eccentricity

• Answer: d
Explanation: Eccentricity, which is the ratio of major
axis to the minor axis which is one of the important
parameter that is used to describe a boundary.
• The term, Curvature is defined as:
a) Rate of change of area
b) Rate of change of slope
c) Slope
d) Rate of change of diameter

• Answer: b
Explanation: Curvature of a boundary is defined as
the rate of change of slope. In general, as the
boundaries tend to be locally ragged, it is difficult to
obtain reliable measures of curvature at a point on a
digital boundary.
• A point ‘p’ is said to be corner point, if the change
of slope is less than 100.
a) True
b) False

• Answer: b
Explanation: In general, a point ‘p’ is said to be
on the straight line segment if the change of slope
is less than 100 and said to be at the corner point if
the change exceeds 900.
• Statistical moments are used to describe the shape of
boundary segments quantitatively.
a) True
b) False

• Answer: a

• Explanation: Statistical moments like mean, variance


and higher order moments can quantitatively describe
the shape of boundary segments.
• Statistical moments is sensitive to rotation.
a) True
b) False
View Answer
• Answer: b
Explanation: The statistical moment technique of
describing the shape of boundary is insensitive of the
rotation of the shape. If desired, size normalization
can be achieved by scaling the range of values of ‘g’
and ‘r’.

You might also like