Image Segmentation with BBO Algorithm
Uploaded by
Editor IJRITCCImage Segmentation with BBO Algorithm
Uploaded by
Editor IJRITCCInternational Journal on Recent and Innovation Trends in Computing and Communication
Volume: 2 Issue: 8
ISSN: 2321-8169
2436 2440
_______________________________________________________________________________________________
Image Segmentation Using Biogeography Based Optimization (BBO)
Sanjeevan Kaur Chahal
Darshan Singh Sidhu
Deptt. Of Electronics and Communication Engg.
PTU Giani Zail Singh Campus
Bathinda,India
sanjeevan_chahal@[Link]
Deptt. Of Electronics and Communication Engg.
PTU Giani Zail Singh Campus
Bathinda,India
[Link]@[Link]
Abstract- Image segmentation is an important problem in computer vision to completely understand the image for better results, i.e., identification
of homogeneous regions in the image and has been the subject of considerable research for over the last three decades. Many algorithms have
been elaborated for this purpose. This paper elaborates two algorithms one is global optimization method Biogeography Based optimization for
automatically grouping the pixels of an color image into disjoint homogeneous regions and the other is clustering method Fuzzy K-means
algorithm for reducing the computational complexity of image. And then comparison between both the techniques is calculated. In this purposed
work these two algorithms are applied to image and performance is evaluated on the basis of computational time. Fuzzy K-means produces
results which require more computational time than Biogeography based optimization. Therefore, comparison shows that Biogeography Based
Optimization is more reliable and faster approach for image segmentation than Fuzzy K-means clustering algorithm.
Keywords- Image Segmentation, BBO, Fuzzy K-means.
_________________________________________________*****_________________________________________________
I.
INTRODUCTION
Segmentation is the process of dividing a digital image into
multiple segments such as a set of pixels, also known as
super pixels. The main objective of segmentation is to
simplify and change the representation of an image into
meaningful image that is more appropriate and easier to
analyze. Segmentation is basically a collection of methods
that allows spatially partitioning close parts of image as
objects [1]. Image segmentation is a very important aspect
of digital image processing. Image segmentation may be
defined as a process of assigning pixels to homogeneous and
disjoint regions which form a partition of the image that
share certain visual characteristics. It basically aims at
dividing an image into subparts based on certain features.
Features could be based on certain boundaries, contour,
color, intensity, or texture pattern, geometric shape or any
other pattern. Image segmentation is used to locate and find
objects and boundaries in image. Region growing is a simple
region based image segmentation method. It is also classified
as a pixel based image segmentation method since it involves
the selection of initial seed points. This approach to
segmentation examines neighboring pixel of initial seed
point and determines whether the pixel neighbors should be
added to the region [1].
Biogeography Based Optimization:
Biogeography Based Optimization is a type of evolutionary
algorithm. As its name implies, BBO is based on
mathematical study of biogeography. Biogeography is the
study of the distribution of animals and plants over time and
space. BBO is an evolutionary process that achieves
information sharing by species migration. It is modeled after
the emigration and immigration of species between habitats
to achieve information sharing. BBO operates by migrating
information between individuals, thus resulting in a
modification of existing individual. Individual do not die at
the end of generation. One such characteristic of BBO is that
original population is not discarded after each generation
rather it is modified by migration [3].
BBO is a population based optimization algorithm it does not
involve reproduction or the generation of offspring. In
1960, the first mathematical equations were discovered and
developed that govern the distribution of organisms.
Therefore, mathematical model of biogeography describe
how species migrate from one island to another, how species
arise and how species become extinct. Biogeography
technique is based basically on two criteria-HSI and LSI.
Geographical area that are well suited and more compatible
residence for biological species are said to have highly
suitability index (HSI). Features that correlate with HSI
include factors such as rainfall, diversity of vegetation and
topographic features, land area and temperature. The
variables that are characterized habitability are called
suitability index variables (SIV). Habitat with HSI tends to
have large number of species, while those with LSI have a
small no of species. HSI are more static than LSI. LSI has a
high species immigration rate because of their scattered
population [2].
2436
IJRITCC | August 2014, Available @ [Link]
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 2 Issue: 8
ISSN: 2321-8169
2436 2440
_______________________________________________________________________________________________
I
Immigration ()
E
Emigration ()
Smax
Number of species
Fig. 1:-Species model of a single habitat containing number
of species
Fig. 1 illustrates a model of species abundance in a single
habitat. The immigration rate and the emigration rate are
functions of the number of species in the habitat. In
immigration curve, the maximum possible rate to the habitat
is I, which occurs when there are zero species in the habitat.
By increasing the number of species the habitat will become
more congested and accordingly fewer species will be able to
adapt immigration to the habitat and therefore immigration
rate decreases. The largest possible number of species in
habitat Smax is at which point the immigration rate becomes
zero. Now if there are no species in the habitat then the
emigration rate of that habitat will be zero. As the number of
species increases more species are able to leave the habitat
and the emigration rate increases. The equilibrium stage
occurs when S occurs when immigration rate and emigration
rate are equal.
exclusive clusters (K) is done by it in such a fashion that
objects in other clusters remain as close as possible to each
other but as far as possible from objects in other clusters.
Each cluster is characterized by its centre point i.e. centroid.
The distances used in clustering in most of the times do not
actually represent the spatial distance. In a dataset, a desired
number of clusters K and set of K initial starting points, the
K- means clustering algorithm find the desired number of
distinct clusters are their centroids. A centroid is a point
whose co-ordinates are obtained by means of computing the
average of each of the co-ordinates of the points of samples
assigned to the clusters [8].
The fuzzy K-means clustering algorithm is a special case of
the generalized fuzzy K-means clustering algorithm scheme,
where point representatives are adopted and the Euclidean
distance is used to measure the dissimilarity between a vector
X and its cluster representative. The fuzzy K-means
clustering (FKM) algorithm performs iteratively the partition
step and new cluster representative generation step until
convergence. This algorithm is one of the most common
methods of clustering which starts with a set of K reference
points and data points belong to K cluster based on distance
criteria [9].
II.
PURPOSED METHODOLOGY
Firstly, image segmentation is done through BBO and then it
is applied to Fuzzy K-means clustering algorithm. So, the
purposed algorithm for both techniques is as discussed
below:Purposed Algorithm for BBO:-
BBO depends on following two theories:Step1:- Take a medical image and convert it into Lab image.
Migration:-The BBO migration property states that whether
to migrate from one region to other or not. The migration
rates of each solution are used to probabilistically share
features between solutions. Migration in BBO is an adaptive
process; it is used to modify existing islands. Migration stage
arises when LSI occurs. When species are less compatible
with their habitat then they migrate (Simon 2008).
Mutation:-The purpose of mutation is to increase habitat
among the population. For low value solutions, mutation
gives them a chance of enhancing the quality of solutions,
and for high fitness value solutions, mutation try to improve
the value as compared to the previous value.
Fuzzy K-means clustering:
Step 2:- Apply optimization technique BBO and clustering
technique on the medical image.
Step 3:- Select the optimum value of population size, fitness
element and immigration rate in BBO. Initialize the BBO
parameters.
Step 4:-Select lambda using BBO in segmentation. Calculate
contour point and fitness value in segmentation.
Step 5:-Calculate elapsed time using BBO technique. Go to
step (3) for next iteration. This loop can be terminated after a
predefined number of generations or after an acceptable
problem solution has been found.
Fuzzy K-means clustering is basically a partitioning method
applied to analyze data and treats observations of the data as
objects based on locations and distance between various
input data points. Partitioning the objects into mutually
2437
IJRITCC | August 2014, Available @ [Link]
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 2 Issue: 8
ISSN: 2321-8169
2436 2440
_______________________________________________________________________________________________
Step 3:-Classification-To examine each point in the dataset
and assigning it to the cluster whose centroid is nearest to it.
Input image (medical or still)
Step 4:-Centroid calculation-When each point in the dataset
is assigned to a cluster, it is needed to recalculate the new k
centroids.
Convert RGB into LAB image
,
Step 5:-Convergence criteria-The steps (3) and (4) are
repeated until no point changes its cluster assignment.
III.
Apply BBO approach and set
probability for each region
RESULTS AND DISCUSSIONS
In this purposed method two algorithms are presented to
reduce the computing time taken to segment the image.
Firstly, medical image is segmented using BBO optimization
technique and then Fuzzy K-means is applied to the same
image. After that two techniques are compared on the basis
of time and then time elapsed to segment the image is
calculated. The purposed work clearly shows that BBO gives
results faster as compared to Fuzzy K-means on the basis of
time taken to segment the image. Following results are
shown with original image and segmented images.
Migrate and mutate the pixels for each
region and compute regions.
Coding (.m file)
Segmented image as output
Fig. 2:-Flowchart of purposed work
Purposed algorithm for Fuzzy K-means:Fuzzy K-means algorithm is one of the most important
clustering algorithm. In this algorithm, the samples are first
divided into two or more clusters and the number of clusters
has already been specified. In Fuzzy K-means clustering
algorithm the main function is:
=
=1 =1
Fig. 3:-Original image of head.
(1)
=1 =1
In formula 1: m is a real number which is bigger than 1
where Xk is the kth sample, Vi is the centre of cluster and n is
the number of samples. Uik shows the dependency of ith
sample in kth cluster. is determined the similarity of
samples (distance) from the centre of the cluster[11].
Step 1:-Set K-To choose a number of desired clusters, K.
Step 2:-Initialization-To choose k starting points which are
used as initial estimates of the cluster centroids. They are
taken as the initial starting values.
Fig. 4:-Segmented image at 70th iterations after applying
BBO algorithm.
2438
IJRITCC | August 2014, Available @ [Link]
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 2 Issue: 8
ISSN: 2321-8169
2436 2440
_______________________________________________________________________________________________
discussed methods. The following comparative table clearly
shows that BBO provides faster results than Fuzzy K-means.
Table I:-Compared time elapsed results using both techniques
Input Images
Fig. 5:-Image labeled by cluster index after applying FKM
algorithm.
Time taken to segment image in
seconds
BBO
Fuzzy K-means
[Link]
4.25
[Link]
3.3
4.12
[Link]
9.83
11.49
IV.
Fig. 6:-Image showing objects in cluster 1
CONCLUSION
As concluded from the results BBO is more reliable and fast
search algorithm for image segmentation purpose. BBO plays
an important role in enhancing the quality and contrast of
natural images. BBO is a better technique to detect abnormal
growth of tissues as compared to other techniques. BBO is a
population based algorithm and it does not involve
reproduction or the generations. Segmentation is a collection
of methods allowing interpreting close parts of the image as
objects. BBO takes less time and provides more accurate
segmented image when compared with Fuzzy K-means
algorithm. So in future, BBO can be used in medical image
segmentation to detect objects. Further image segmentation
techniques or noise removal methods can be improved so that
the input image to be extracted could be made more efficient
which can improve the final outcome.
REFERENCES
[1]
[2]
Fig. 7:-Image showing objects in cluster 2
[3]
Figure.3 shows the original image of head and figure.4 shows
segmented images of head after applying BBO technique at
70th iteration. Figure.5 shows the cluster index of image after
applying FKM algorithm. Figure.6 and 7 shows the
segmented image in cluster 1 and 2 respectively. To further
evaluate the versatility of purposed method, different medical
images are taken as example and are subjected to above
[4]
[5]
Surbhi Gupta, Krishma Bhuchar, Parvinder [Link],
Implementing Color Image Segmentation Using
Biogeography Based Optimization, International
Conference on Software and Computer Applications
IPCSIT vol.9, pp.72-78, 2011.
Rajwinder Kaur, Rakesh Khanna, Medical Image
Quantization using Biogeography Based Optimization,
International Journal of Computer Applications(09758887), vol.48, No.12, pp.8-11, 2012.
Dan Simon, Biogeography Based Optimization IEEE
Transaction on Evolutionary Computation, vol.4, No.6,
pp.702-713, 2008.
Mittu Mittal, Gagandeep, Comparison between BBO and
Genetic Algorithm, International Journal of Science,
Engineering and Technology Research (7278-7798),
vol.2, No.2, pp.284-294, 2013.
Haping Ma, Suhang Ni and Man Sun, Equillibrium
Species Counts and Migration Model Tradeoffs for
Biogeography Based Optimization, Joint 48th IEEE
2439
IJRITCC | August 2014, Available @ [Link]
_______________________________________________________________________________________
International Journal on Recent and Innovation Trends in Computing and Communication
Volume: 2 Issue: 8
ISSN: 2321-8169
2436 2440
_______________________________________________________________________________________________
[6]
[7]
[8]
[9]
[10]
conference on decision and control and 28th Chinese
control conference, pp.3306-33010, 2009.
S.V.S Prasad, Satya T. Savitri, Krishna I.V. Murali,
Classification of Multispectral Satellite Images using
Clustering with SVM Classifier, International Journal of
Computer Application(0975-8887), vol.35, No.5, pp. 702713, 2011.
Krishna Bhuchar, Rekha Rani, Bharti Jyoti, Performance
Evaluation of Biogeography Based Image Segmentation,
UACEE International Journal of Computer Science and its
Applications(2250-3765), vol.2, No.1, pp.47-51.
Saumi Ghosh, Sanjay Kumar Dubey, Comparitive
Analysis of Fuzzy K-means and Fuzzy C-means
Algorithms, International Journal of Advanced Computer
Science and Applications, vol.4, No.4, pp.35-39, 2013.
Chih-Tang Chang, [Link] and Der-Mu Jeng, A
Fuzzy K-means Clustering Algorithm using Cluster
Center Displacement, Journal of Information Science and
Engineering.27, pp.995-1009, 2011.
Farhad Soleimanian Gharechopogh, Neha Jabbari, Zeinab
Ghaffari Azar, Evaluation of Fuzzy K-means Clustering
Algorithms in Intrusion Detection System, International
Journal of scientific and technology research, vol.1,
No.11. pp.66-72, 2012.
2440
IJRITCC | August 2014, Available @ [Link]
_______________________________________________________________________________________
Common questions
Powered by AIThe initialization process for BBO and Fuzzy K-means in image segmentation diverges significantly in their approaches to setting starting conditions. BBO initializes by selecting optimal values for parameters such as population size and immigration rate, which guide the solution migration and mutation processes without discarding the existing population after each generation. In contrast, Fuzzy K-means initializes by selecting a predefined number of clusters (K) and starting points that serve as initial estimates for cluster centroids. This involves partitioning the data into clusters and rotating through cluster assignment and centroid recalculation until convergence, which is more resource-intensive than BBO's approach .
When applied to the same image dataset, BBO demonstrates superior computational efficiency compared to Fuzzy K-means. The BBO algorithm, by leveraging the concepts of habitat suitability and species migration, processes images significantly faster as it applies a collective population-based search rather than recalculating cluster centroids iteratively as in Fuzzy K-means. This efficiency is illustrated in comparative studies, showing BBO consistently segments images quicker, reducing computational time across multiple test cases .
Species migration in BBO allows for effective information sharing between possible solutions, enhancing the optimization process in image segmentation. This migration mirrors the emigration and immigration processes seen in nature, permitting different 'species' or solution components to move between 'habitats' or potential solutions. This dynamic sharing of information helps refine the population of solutions iteratively, which aligns with exploring multiple possible segmentations efficiently, leading to a more robust identification of homogeneous image regions .
Potential future applications of BBO in image segmentation inferred from its current uses include more sophisticated real-time processing in dynamic imaging systems, such as medical diagnostics that require immediate analysis of imaging data. BBO's rapid convergence and adaptability could also enhance automated surveillance systems, where fast and accurate recognition of object patterns is critical. Additionally, its adaptability to identify subtle features could benefit astronomical image processing, detecting faint celestial bodies and phenomena that rely on precise segmentation algorithms .
The High Suitability Index (HSI) and Low Suitability Index (LSI) have crucial implications on species migration within BBO by dictating the conditions under which solutions are likely to migrate. Habitats with high HSI, which are conducive environments for species, have a low rate of immigration due to being near optimal already, whereas those with LSI have higher immigration rates due to needing more improvement. This migration of solutions allows BBO to dynamically adjust and optimize solutions by probabilistically sharing features between solutions that are less adapted to highly adapted ones, effectively guiding the process of image segmentation towards optimal solutions .
Biogeography-Based Optimization (BBO) is advantageous compared to Fuzzy K-means for image segmentation mainly due to its faster computational time and reliable performance. BBO, being a global optimization method inspired by the mathematical study of biogeography, efficiently groups pixels into homogeneous regions through species migration. Unlike Fuzzy K-means, which requires more computational time due to its clustering process that analyzes data based on location and distance, BBO simply modifies the existing population without reproduction, contributing to its quicker processing .
The adaptability of BBO's migration and mutation processes improves image segmentation outcomes by allowing solutions to dynamically adjust based on the evolving needs of the problem space. Migration allows for the exchange of beneficial solution traits between habitats, effectively spreading advantageous features to less suitable habitats, while mutation introduces necessary variability, providing a mechanism for continuous solution improvement. Together, these processes ensure a thorough exploration and refinement of potential segmentations, leading to more accurate and robust identification of image objects and boundaries .
Mutation in the BBO algorithm serves to enhance solution variability and quality by introducing small random changes to the habitat attributes, which represent potential solutions. For low-quality solutions, mutation provides a chance to improve by moving closer to higher fitness values, expanding the search space and preventing premature convergence. For high-quality solutions, mutation ensures they are not stuck in local optima by exploring potential improvements. Thus, mutation contributes to maintaining diversity in solution populations, helping achieve better segmentation results in image processing tasks .
Converting an RGB image to a Lab image before applying BBO for segmentation is significant because the Lab color space separates the lightness component from chromatic components, allowing for more accurate and perceptually uniform color differentiation. This conversion facilitates sophisticated processing in BBO, enabling it to better isolate image features based on lighting and color distinctiveness, thus improving the effectiveness of the segmentation process by more accurately delineating homogeneous regions .
BBO's effectiveness in detecting abnormal tissue growth in medical images over Fuzzy K-means stems from its ability to precisely optimize image segments through global search mechanisms and information sharing between solutions. BBO enhances the quality and contrast of images by efficiently exploiting the entire solution space for segmentation, thus accurately identifying subtle anomalies and object boundaries that might be missed by Fuzzy K-means, which may suffer from local minima due to its centroid-based clustering approach. This precision is critical in medical imaging where accurate tissue characterization is necessary .