P a g e | 940

International Journal of Research (IJR) Vol-1, Issue-7, August 2014 ISSN 2348-6848
A Review for Object Recognition using Color Based Image Quantization
Technique | Manpreet kaur, Silky Narang
A Review for Object Recognition using Color Based
Image Quantization Technique

Manpreet Kaur
1
, Silky Narang
2

1 Student, Dept. of Information Technology, Adesh Institute of Engineering and Technology
Under Punjab Technical University , Jalandhar- Kapurthala Highway,
Near Pushpa Gujral Science City,
Kapurthala, Punjab 144601,India
2 Faculty, Dept. of Information Technology, Adesh Institute of Engineering and Technology
Under Punjab Technical University , Jalandhar- Kapurthala Highway,
Near Pushpa Gujral Science City,
Kapurthala, Punjab 144601,India

ABSTRACT
This paper deals with a review on various
Color Quantization Techniques .The aim is to
propose Exemplar-based object recognition
which uses an adaptive color quantization
scheme to obtain a coarse image representation.
The tiny regions are combined based on color
information. The Energy transform function
using extracted color map is used as a criterion
for image segmentation. The motivation for the
proposed method is to obtain the similar and
significant objects in different images using
color quantization.
Key words-Color Quantization, color map,
Fuzzy c-means clustering algorithm, kernel-
based Fuzzy c-means clustering algorithm

1. INTRODUCTION

Color quantization is critical for
displaying images with many colors on devices
that can only display a limited number of colors,
usually due to memory limitations, and enables
efficient compression of certain types of images.
A color image quantization is a process
that reduces the number of distinct colors used
in an image, usually with the intention that the
new image should be as visually similar as
possible to the original image.
The process of color image quantization
is often broken into four phases, Heckbert[2].
Phase 1 is sampling the original image for
color statistics.
Phase 2 is choosing a color map based on
those statistics.
Phase 3 is mapping the colors to their
representative in the color map.
Phase 4 is Quantizing and drawing the
new image.
Phase 4 is a trivial matter regardless of
the quantization method. The other three phases
however are more strongly connected. In
particular the method used for phases 1 and 2
will determine the best method for
accomplishing phase 3.
In general algorithms for color
quantization can be broken into two categories:
Uniform and Non-Uniform.
Uniform: Here the color space is broken
into equal sized regions where the number of
regions, NR is less than or equal to K.
Non-Uniform: Here the manner in which
the color space is divided is dependent on the
distribution of colors in the image


P a g e | 941
International Journal of Research (IJR) Vol-1, Issue-7, August 2014 ISSN 2348-6848
A Review for Object Recognition using Color Based Image Quantization
Technique | Manpreet kaur, Silky Narang
2. RELATED WORK [3]
Several heuristic techniques for color
image quantization have been proposed in the
literature. Some of them are discussed below.
The median cut algorithm (MCA) divides
the color space repeatedly along the median into
rectangular boxes until the desired number of
colors is obtained.
Popularity Algorithm builds the color
map by finding the K most frequently appearing
colors in the original image. Therefore the
colors are stored in a histogram. The K most
frequently occurring colors are extracted and
they are made the entries in the color table. Now
the true image can be quantized.
The variance-based algorithm (VBA) also
divides the color space into rectangular boxes.
However, in VBA the box with the largest mean
squared error between the colors in the box and
their mean is split.
The octree quantization algorithm
repeatedly subdivides a cube into eight smaller
cubes in a tree structure of degree eight. Then
adjacent cubes with the least number of pixels
are merged. This is repeated until the required
number of colors is obtained.
M. G. Omran in his paper proposes Color
image quantization based on PSO. The proposed
approach is of the class of quantization
techniques that performs clustering of the color
space. The proposed algorithm randomly
initializes each particle in the swarm to contain
K centroids (i.e. color triplets). The K-means
clustering algorithm is then applied to each
particle at a user-specified probability to refine
the chosen centroids. Each pixel is then assigned
to the cluster with the closest centroid. The PSO
is then applied to refine the centroids obtained
from the K means algorithm.
The Recent Clustering applied in color
image quantization is the c-mean clustering
algorithms, which is a very time consuming
approach. To reduce the computing load FCM
algorithm based on color quantization is
effective and fast to extract object from
complicated Background but unable to have
better curve Evolution.



3.FCM algorithm based on color
Quantization Technique

The Fuzzy c-means clustering algorithm
was first introduced by Dunn and later extended
by Bezdek[11].Bezdek proved convergent
property of clustering algorithm and hard C-
means clustering algorithm. Here after, an
objective-function-based fuzzy C-means
clustering algorithm, has been extensively
applied in such fields as medical diagnosis ,
target identification and image segmentation
etc.The algorithm partition the data set
X={x1,x2,x3,.,xn}R
pm
into c class with the
result U={uik} R
cn
,where uik is the degree of
membership of xk belonging to ith the cluster. In
this algorithm, every image pixel must be used
for iterative calculation load is heavy. With
color quantization, we used the representative
color of each subset for FCM clustering,
improving the running speed greatly. The data
set in the quantized image is denoted as(Q,H),
where
Q={x1,x2,x3,.,xq}R
pq
,
H={h(1),h(2),.,h(q)},
h(k) is the number of pixels in the k
th
subset with
the representative color xk. The fuzzy c-means
(FCM) clustering algorithm is an iterative
clustering method that produces an optimal
partition by minimizing the objective function.
[
m
(u, I) = (u
k
)
m
J
k
2
(x
k
, :

)
c
=1
n
k=1

.. (1)
By color quantization, objective function
become
[
m
(u, I) =
b(k)(u
k
)
m
J
k
2
(x
k
, :

)
c
=1
q
k=1

.. (2)
Where U={uik}is define as
u
k
= _ _
d
ik
(x
k
,
i
)
d
]k
(x
k
,
]
)
]
2
(m-1)
,
c
]-1
_
-1
, i =
1.. . c k = 1.. q. (3)
: = {:
1
, :
2
... . :
c
] e R
pc
are c cluster
centres computed with
:

=
(u
ik
)
m
x
k
n
k=1
(u
ik
)
m n
k=1
. (4)
With color quantization, formula (4) is
replace with
:

=
h(k)(u
ik
)
m
x
k
q
k=1
h(k)(u
ik
)
m n
k=1
. (5)

P a g e | 942
International Journal of Research (IJR) Vol-1, Issue-7, August 2014 ISSN 2348-6848
A Review for Object Recognition using Color Based Image Quantization
Technique | Manpreet kaur, Silky Narang
From mentioned above, because q is
much smaller than n, the calculation load of
objective function ,membership and centres are
reduced greatly and the whole running speed of
FCM is thus much improved .After image pre-
processing ,pixels are distributed more uniform
and the data clustering ,could be performed
directly in RGB space without affecting the
clustering result. So, the proposed CQ-FCM
could greatly improve the running speed without
reducing quality of image segmentation.[6]

4. CONCLUSION

There are many different algorithms for
color quantization. Which one is best is
dependent on the requirements of the system it is
to be used in. The uniform quantization
algorithms offer a quick and dirty means but can
result in very poor quality depending on image
characteristics. The non-uniform algorithms
offer better consistent results at the cost of
increased memory and time complexities. To
have a exemplar based object recognition in an
images we propose Color image quantization
with KFCM[5]. Color image quantization has
been applied using bacteria foraging method.
Bacteria Foraging Optimization Algorithm
(BFOA), proposed by Passino, is a new comer to
the family of nature inspired optimization
algorithms. After obtaining color approximated
image according to synthetic color map, kernel
based FCM has been applied in order to carry
out segmentaion process. The main motives of
using the kernel methods consist in: (1) inducing
a class of robust non-Euclidean distance
measures for the original data space to derive
new objective functions and thus clustering the
non- Euclidean structures in data; (2) enhancing
robustness of the original clustering algorithms
to noise and outliers, and (3) still retaining
computational simplicity. The algorithm is
realized by modifying the objective function in
the conventional fuzzy c-means (FCM)
algorithm using a kernel induced distance
instead of Euclidean distance in the FCM, and
thus the corresponding algorithm is derived and
called as the kernelized fuzzy c-means (KFCM)
algorithm, which is more robust than FCM.

5. ACKNOWLEDGEMENTS
This study was conducted by the rst author
under the supervision of the co-author in partial
fulllment of the requirements of a Master
Degree in Information Technology. The First
author wishes to thank Assistant Professor Silky
Narang at Adesh Institute of Engineering and
Technology, Faridkot under Punjab Technical
University , Jalandhar for her support over the
period in which this article was written.
6. REFERENCES

1. Gervautz, M. & Purgathofer, W. "A
Simple Method for Color Quantization:
Octree Quantization", Graphics Gems,
Academic Press, 1990
2. Heckbert, P. "Color Image Quantization
for Frame Buffer Display", Computer
Graphics, Vol 16, #3, pp. 297-303, 1982
3. Heena and Aggarwal, H. , "Color Image
Quantization Based on Euclidean
Distance Using Bacteria Foraging
Optimization," International Journal of
Electronics and Computer Science
Engineering, ISSN 2277-1956/V1N4-
2285-2290
4. Hill, F. S. Jr. "Computer Graphics",
Macmillna Publishing Co., 1990
5. Hogo, Mofreh A. (2010) Evaluation of
E-learners Behaviour Using Different
Fuzzy Clustering Models: A
Comparative Study IJCSIS Vol. 7, No.
2.
6. Lin Kaiyan, Wu Junhui,Chen ji,Shi
huping A Real Time Image
segmentation Approach for Crop Leaf
,IEEE 2013
7. Wu, Xiaolin "Color Quantization by
Dynamic Programming and Principle
Analysis", ACM Transactions on
Graphics, Vol. 11, No. 4, pp. 348-372,
1992
8. Wu, Xiaolin "Efficient Statistical
Computations for Optimal Color
Quantization", Graphics Gems II,
Academic Press, 1991