University Mustapha Stambouli Mascara 2018-2019
Feature Extraction state of the art
Presented by: Mohamed Amine Chadli.
Encadreur : Dr. Bachir bouiadjra Rochdi
Co-encadreur : Dr. Fekir Abdelkader
1
University Mustapha Stambouli Mascara 2018-2019
Summary
●
What are moments?
●
Definition
●
Geometric moments
●
Moments in images
●
Some meaning of moments
●
Invariant moments
●
Translation
●
Rotation
●
Scaling
●
Moments invariants in detail
2
●
The central moments
What are moments?
●
Moments are scalar quantities used to
characterize a function and to capture its
significant features.
3
Definition
●
General moment Mpq of an image
f(x, y), where p, q are non-negative
integers and r = p + q is called the order of the
moment, defined as:
4
Definition
●
where p00(x, y), p10(x, y), . . . ,
pkj(x, y), . . . are polynomial basis functions
defined on D.
5
Geometric moments
●
The most common choice is a standard power
basis
pkj(x, y) = x^k * y^j that leads to geometric
moments
6
Moments in images
●
If the image is represented by a discrete
function, integrals are replaced by
summations. Equation above can be written
as follows:
7
Some meaning of moments
●
Geometric moments of low orders have an
intuitive meaning:
– m00 is a “mass” of the image (for binary images, m00 is an area of the object)
– Centroid:
– Second-order moments m20 and m02 describe the “distribution of mass” of
the image with respect to the coordinate axes
8
Invariant moments
●
Invariant moments are features invariant to:
– Translation
– Rotation
– Scaling
9
Translation
●
Invariant to translation is when the feature descriptor retrieve
the same information even if the shape of the object in the
image is shifted from its former place. For example:
10
Rotation
●
Invariant to rotation is when the feature descriptor
retrieve the same information even if the the shape of
the object is rotated for (n degree)
11
Rotation
12
Scaling
●
Invariant to scaling is when the feature descriptor retrieve the
same information even if the the shape of the object is scaled
up or down as shown below:
13
Moments invariants in detail
●
The famous invariants moments are:
– The central moments
– The normalized central moments
– Hu invariant moments
14
The central moments
●
The central moments of an image
f(x, y) are defined by:
–
– Where are the centroid of
the image.
– The central moments are invariant to translation.
15
The central moments
●
The central moments of order up to 3 are as follows:
16
The normalized central moments
●
The normalized central moment to shape and size of
order (p+q) is defined as:
where and p + q
– This moments are invariant to both translation and
scale.
17
Hu invariant moments
●
In 1962 Hu were the first who introduced
moments invariants to each translation, scale
and rotation.
●
Based on the normalized central moments Hu
derived seven moment invariants.
18
Hu invariant moments
19
University Mustapha Stambouli Mascara 2018-2019
Conclusion
Moments invariants are so important in the field of
features extractions therefore understanding how they
work is a must.
In the next work we will try to implement them to
understand them better.
20
University Mustapha Stambouli Mascara 2018-2019
References
[1] Moments and Moment Invariants in Pattern Recognition; Jan Flusser, Tomáš Suk, Barbara
Zitová; 2009
[2] Feature Extraction Based on Moment Invariants for Handwriting Recognition; S. C. Mehrotra,
R. J. Ramteke;
21
