PERFORMANCE ANALYSIS USING SINGLE SEEDED REGION GROWING ALGORITHM

International Journal of Innovative Research in Advanced Engineering (IJIRAE) ISSN: 2349-2163
Volume 1 Issue 6 (July 2014) http://ijirae.com
_________________________________________________________________________________________________
© 2014, IJIRAE- All Rights Reserved Page - 224
PERFORMANCE ANALYSIS USING SINGLE SEEDED
REGION GROWING ALGORITHM
Shilpa Dantulwar (Kamdi)1 Prof. R.K.Krishna2
M Tech IV Semester Computer Science & Dept. Of Electronics, Rajiv Gandhi College
Engineering, RCERT, Chandrapur of Engg., Res. &Tech, Chandrapur(M.S)
-------------------------------------------------------------------------------------------------------------------------------------
A b s t r a c t : Image segmentation is an important process and its results are used in many image processing
applications. Color images can increase the quality of segmentation, but increase the complexity of the problem. This
paper focuses on measurement of parameters that is RI,GCE,MMSE and time for segmentation using "Seeded Region
growing algorithm". Image segmentation techniques using region growing requires initial seeds selection, which
increases computational cost & execution time. To overcome this problem, a single seeded region growing technique for
image segmentation is proposed, which starts from the center pixel of the image as the initial seed. It grows region
according to the grow formula and selects the next seed from connected pixel of the region. The optimization is done
with fuzzy logic to improve the value of parameters.
Keywords: Image segmentation, fuzzy logic, Single seeded region growing algorithm, Grow formula , Connected pixel,
Random Index.
---------------------------------------------------------------------------------------------------------------------------------------------------
1. INTRODUCTION
Seeded region growing algorithm (SRG) is a new approach which is based on conventional postulate of region growing
algorithms where the criteria of similarity of pixels is applied, but the mechanism of growing regions is closer to the
watershed algorithm. Instead of controlling region growing by tuning homogeneity parameters, SRG is controlled by
choosing a usually small number of pixels, known as seeds. These seed pixels are chosen by user according to his own
opinion what should be regions to extract on the image[26].
This paper focuses on a new method for image segmentation and its improvements. This method is called "Seeded Region
Growing" and it is introduced by Rolf Adams and Leanne Bischof[21].They proposed a new method for segmentation of
intensity images, which is robust, rapid and free of tuning parameters. These characteristics allow implementation of one
very good algorithm which could be applied to large variety of images. This method, however, requires selection of seed
regions, what has to be done manually and it classifies this approach to the class of semiautomatic algorithms. The
algorithm grows these seed regions until all of image pixels have been processed. Unfortunately, the algorithm is
dependent on the order of pixel processing, what means that, for example, raster order and anti-raster order processing of
the pixels on the image will lead to the different regions on the image. This problem is revisited by Andrew Mehnert and
Paul Jackway in their improved version of algorithm. In this approach the process of segmentation is completely
automatic, which is needed in many applications, as one of main disadvantages appears manual selection of the seed pixels
or regions. This problem of manual selection of seed pixels or regions could be solved by automatic selection of the seed
pixels. This paper is organised as follows; section 2 introduces the classical approaches for the segmentation of intensity
images, section 3 introduce the basic of seed selection; Section4 fuzzy logic,section 5 discuss proposed algorithm work and
measurement of parameters,section 6 shows result and we draw some conclusion in section 7.
II. CLASSICAL APPROACHES FOR THE SEGMENTATION OF INTENSITY IMAGES
There are three methods to classify intensity images
A. ThresholdTechniques
Threshold techniques are based on the thresholds which are usually selected from the image histogram. It is said
that all pixels whose value (gray intensity, color, or other) is between two values of thresholds belong to one
region. The fact that the thresholds are derived from the histogram .this method have problems to cope well with
the noise as well as with blurred edges on the image.
B. Edge-BasedMethods
Edge based methods try to find the places of rapid transition from one to the other region of different brightness or
color value. The basic principle is to apply some of the gradient operators convolving them with the image. High
values of the gradient magnitude are possible places of rapid transition between two different regions, what we
call edges. After this step of finding edges on the image they have to be linked to form closed boundaries of the
regions.
C. Region-BasedMethods
Region-based methods are complementary to the edge-based methods. Here the point is to group pixels of the

_________________________________________________________________________________________________
same or similar brightness or color to the regions according to the given criteria of homogeneity. Homogeneity
criteria is based on some threshold value, the choice of which is problematic, because we have usually to play a
lot with the right choice of the thresholds, and thresholds always depend on the image data. Also, results are
sensitive to the noise.
D. Mixed, Hybrid Methods Which Combine Edge And Region Methods
There are a lot definitions of such mixed methods. . Comparing with edge-based methods this method gives closed
boundaries around the supposed regions, but it is sensitive to the noise in the regions and blurred edges between
regions. The method is highly data driven with no scope to involve higher level knowledge. It is computationally very
region growing method based on variable-order surface fitting. The method is highly data driven with no scope to
involve higher level knowledge. It is computationally very expensive.
III. Basic of Seed Selection Method
The first step in region growing is to select a set of seed points. Seed point selection is based on some user
criterion (for example, pixels in a certain gray-level range, pixels evenly spaced on a grid, etc.). The initial region begins as
the exact location of these seeds.The regions are then grown from these seed points to adjacent points depending on a
region membership criterion. The criterion could be, for example, pixel intensity, gray level texture, or color.
Since the regions are grown on the basis of the criterion, the image information itself is important. For example, if
the criterion were a pixel intensity threshold value, knowledge of the histogram of the image would be of use, as one could
use it to determine a suitable threshold value for the region membership criterion[25].
A. Some important issues:
Then we can conclude several important issues about region growing are:
1. The suitable selection of seed points is important.
2. More information of the image is better.
3. The value, “minimum area threshold”.
4. The value, “Similarity threshold value“.
B. Properties of SRG algorithm
SRG method may not be applied to high textured images or to range images. It may be applied to images affected by
light variation but only after initial preprocessing (i.e. removal of the background). SRG also can be extended for color and
multispectral images. This method can be implemented on any number of dimensions as well as on any shape grid (or
graph). Two very important problems which segmentation algorithm should overcome are:
 robustness in the presence of unwanted small round object on the image
 robustness in the presence of noise
SRG algorithm doesn't depend on position where initial seed pixels are put. This choice is left to the end-user. Evaluation
of the segmentation results is also made by user. This fact classifies SRG to the group of semiautomatic algorithms. It is
possible that user makes mistake selecting seed pixels. That can lead to less or more regions on the images. SRG algorithm
is semi automated, but is it possible to automatize the whole process of segmentation? Generally it is possible, but it
depends on what is your goal and on which images you want to apply that. Usually you can start with some simple method
of segmentation, for example, based on thresholding. Than you can use results of this segmentation as starting point for
SRG.
IV. Fuzzy Logic
Fuzzy systems are mappings from the family of fuzzy sets to the family of fuzzy sets. In other words, they are operators
transforming fuzzy sets to fuzzy sets. The fuzzy sets may be onedimensional as well as multidimensional.In practice, often
the input of the fuzzy system as well as the output are crisp numbers. But a fuzzy system requires a fuzzy set as input and
generates a fuzzy set as output. Therefore schemes
like the one displayed are normally used.
Fig 1. Fuzzy Logy
A. Advantages and disadvantages of the use of fuzzy systems
i. Advantages:
 Fuzzy logic sometimes uses only approximate data, so simple sensors can be used.

_________________________________________________________________________________________________
 The algorithms can be described with little data, so little memory is required.
 The algorithms are often quite understandable.
 Fuzzy algorithms are often robust, in the sense that they are not very sensitive to changing environments and
erroneous or forgotten rules.
ii. Disadvantages:
 In areas that have good mathematical descriptions and solutions, the use of fuzzy logic most often may be sensible
when computing power (i.e. time and memory) restrictions are too severe for a complete mathematical
implementation.
 Proof of characteristics of fuzzy systems is difficult or impossible in most cases because of lacking mathematical
descriptions.
V. Proposed Algorithm
After studying the various approaches we find that there is no unique method or approach for image segmentation.
Conventional image segmentation techniques using region growing requires initial seeds selection, which increases
computational cost & execution time.
To overcome this problem,and to calculate values of stastical parameter like MMSE and Time, a single seeded region
growing technique for image segmentation is proposed, which starts from the center pixel of the image as the initial seed. It
grows region according to the grow formula and selects the next seed from connected pixel of the region.Then the result of
segmentation is optimized using fuzzy logic
Brief review of region growing
A. Seeded Region Growing
Basic function of region growing technique is partition of an image into non overlapped regions. It takes seed.s as input,
and then merge pixels with similar property and produce a region correspond to each seed. Result of region growing must
follow following constraints:
1
Here L is total no. of regions. It means, all regions should form whole image.
2 Ri is connected region, i = 1, 2, 3…n, where n is the number of regions.
3 mutual exclusion of region.
To perform region growing we need to address following
steps:
1. Selection of initial seeds
Selection of initial seeds plays a prominent role in the process of image segmentation. Seeds should have some similar
feature with respect to their neighbors. There should be a seed for every expected region in image. No seeds should be
connected to each other.
2. Growing formula based on stopping criterion
Growing formula decides the homogeneity between seed and neighbors of it based on similarity index. Stopping criteria
should be efficient to discriminate neighbor elements in non homogeneous domain.
B Otsu’s Method For Adaptive Threshold
In 1979, N. Otsu proposed the maximum classes’ variance method (known as the Otsu method) . For its simple calculation,
stability and effectiveness it has been widely used.It is a well-behaved automatic threshold selection method, and its
consumed time is significantly less than other thresholding algorithms. This method regards the largest inter-class variance
between target and background as a principle to choose the best segmentation threshold. Otsu method chooses the optimal
threshold t by maximizing the between-class variance, which is equivalent to minimizing the within-class variance, since
the total variance (the sum of the within-class variance and the between-class variance) is constant for different partitions.
In the proposed single seeded region growing algorithm Seed selection is the first step of the region growing
technique. Instead of selecting seeds initially we select center pixel of the image as initial seed.For the grow formula we
use the intensity based similarity index between the seed and the 8-neighbour pixels of the seed.The similarity index
between two neighborhood pixels of RGB intensities (x, y) and (x+i, y+j) respectively is calculated by Euclidean distance
as follows:

_________________________________________________________________________________________________
The stopping criterion for the grow formula is determined from the Otsu’s adaptive thresholding method. If the distance
between labeled pixel and non labeled pixel is less than the threshold, then we label both pixels as belonging to same
region. In the proposed algorithm we input no of pixel to be grown and the maximum color difference with this algo check
the number of pixel which is given as input from center point and the pixel whose threshold is whine the maximum color
difference is belonging to region otherwise it considered to be belonging to next region.
C Optimization Of SSRGA Using Fuzzy Logic.
For optimization we are using Fuzzy interference system where we select two input that is region and color and output is
area. With the given input one fuzzy threshold is calculated and pixel whose intensity or threshold is whine the range is
belonging to region otherwise it form next group. Same parameter is calculated with optimized algo then we find that
MMSE is slightly increase but time for segmentation is reduced much. We use the concept of Quantization and
aggregation, in which input imge is first quantize and then aggregate to its nearest value. Quantize image then apply as
input to the optimized algorithm.
Fig 2. FIS Editor
VI. RESULTS:
The proposed algorithms have been implemented using MATLAB. The performance of various image segmentation
approaches are analyzed and discussed. The measurement of image segmentation is difficult to measure. There is no
common algorithm for the image segmentation. The statistical measurements could be used to measure the quality of the
image segmentation .The following parameters are used to measure the quality of the image segmentation. The rand index
(RI), global consistency error (GCE), MMSE and time for segmentation. The detailed description of parameters are
explained in detail as follows,

_________________________________________________________________________________________________
i. Minimum mean square error(MMSE): Minimum mean square error is the average of the difference
between predicted and actual value . MMSE indicates that higher the values of error mean the image is of poor
quality. MMSE=mse_num/(mse_den*N)
ii. RI(Random index):
RI counts the fraction of pairs of pixels whose labeling are consistent between the computed segmentation and the
ground truth averaging across multiple ground truth segmentations. The Rand index or Rand measure is a measure of
the similarity between two data clusters.
 a, the number of pairs of elements in S that are in the same set in X and in the same set in Y
 b, the number of pairs of elements in S that are in different sets in X and in different sets in Y
 c, the number of pairs of elements in S that are in the same set in X and in different sets in Y
 d, the number of pairs of elements in S that are in different sets in X and in the same set in Y
The Rand index (RI) is, RI= a+b /(a+b+c+d)
iii. GCE( Global consistency error):
GCE measures the extent to which one segmentation can be viewed as a refinement of the other. Segmentations
which are related are considered to be consistent, since they could represent the same image segmented at
different scales
Output for SSRGA, OSSRGA and QOSSRGA
Enter growing area:8
Enter maximum color difference:20
Table 1.Parameters values for 3 images
(a) (b) (c) (d)
Fig 3.Output for image 368078
(a) (b) (c) (d)
Fig 4. Output for image 118020
(a) (b) (c) (d)
Fig 5. Output for image 118035

_________________________________________________________________________________________________
Methods RI MMSE GCE
QOSSRGA 0.9687 14.68 0.0091
OSSRGA 0.9687 15.78 0.0091
SSRGA 0.9687 14.75 0.0091
Table 2. Average Performance Calculation
Fig 5. Performance Analysis
SSRGA OSSRGA QOSSRGA
149.32 S 45.9366 S 32.1866 S
Table 3. Average Time Analysis
Fig 5. Average Time Analysis
VII. CONCLUSION
A new approach to segment an image using a single seed based region growing algorithm has been proposed in this paper.
In this method the center pixel of the image is selected as the initial seed and the region is grown according to growing
formula with the stopping criterion determined by Otsu’s. Parameters values are calculated using proposed algorithm and
optimized algorithm, we found that with optimized algorithm the value for MMSE is slightly increase but time for

_________________________________________________________________________________________________
segmentation is decrease more than half. When image quantized and aggregated and given as input to optimize algorithm,
we found that MMSE is reduced and time for segmentation is also reduced.
REFERENCES
[1] Indra Kanta Maitra et al, “Detection of Abnormal Masses using Divide and Conquer Algorithmin Digital Mammogram”. International Journal of
Emerging Sciences, 1(4), 767-786, December 2011 (IJES).
[2] Harikrishna Rai et al, “ Gradient Based Seeded Region Grow method for CT Angiographic Image Segmentation.” InterJRI Computer Science and
Networking, Vol. 1, Issue 1, July 2009
[3] Yuxi Chen, Chongzhao Han, “ A hybrid method of seeded region growing and region hue-area information fusion for object segmentation under
patterned background”
[4] Sevasti-Zoi Karakozoglou et al, “Automatic glottal segmentation using local-based active contours and application to glottovibrography.” Speech
Communication xxx (2011) xxx–xxx.
[5] Aman avazpour et al , “Segmentation of extrapulmonary tuberculosis infection using modified automatic seeded region growing”.Published online
2009 July 14. doi: 10.1007/s12575-009-9013-0.
[6] Jia-Hong Yang, “Anisotropic Diffusion with MorphologicalReconstruction and Automatic Seeded Region Growing for Color Image Segmentation”
2008 International Symposium on Information Science and Engieering .
[7] Jie Wu, “Texture Feature based Automated Seeded Region Growing in Abdominal MRI Segmentation” 2008 International Conference on
BioMedical Engineering and Informatics.
[8] Frank Y. Shih et al, “Automatic seeded region growing for color image segmentation” Image and Vision Computing Copyright © 2012 Elsevier B.V.
[9] Jianping Fan et al. “Automatic Image Segmentation by Integrating Color-Edge Extraction And Seeded Region Growing”.IEEE Trans. On Image
Processing(2001).
[10] Li, Weiwei1 et al, “ A color image segmentation method based on automatic seeded region growing”. IEEE International Conference on Automation
and Logistics, ICAL 2007.
[11] Xu, Guoxiong et al, “A color image segmentation algorithm by integrating watershed with automatic seeded region growing and merging”. SPIE -
The International Society for Optical Engineering-2011.
[12]He, Yuan, “Automatic seeded region growing based on gradient vector flow for color image segmentation”. International Journal:IOSR JOURNALS-
2007.
[13]Octavio Gómez et al, “Image Segmentation Using Automatic Seeded Region Growing and Instance-Based Learning”. International Journal:IOSR
JOURNALS[14]Chung-Chia Kang Wen-June Wang, “ Fuzzy based seeded region growing for image segmentation” NAFIPS 2009.
[15] Muhammad Khusairi Osman, “Combining Threshoding and Clustering Techniques for Mycobacterium tuberculosis Segmentation in Tissue
Sections”. Australian Journal of Basic and Applied Sciences, 5(12): 1270-1279, 2011.
[16] M. M. Abdelsamea, “An Automatic Seeded Region Growing for 2D Biomedical Image”, 2011 International Conference on Environment and
BioScience.
[17] N. A. Mat-Isa, “Seeded Region Growing Features Extraction Algorithm;Its Potential Use in Improving Screening for Cervical Cancer”. International
Journal of The Computer, the Internet and Management Vol. 13#1 (January – April, 2005)
[18] Wesolkowski, S., and P. Fieguth, “Color image segmentation using a region growing method”. 2001 Advanced Imaging Conference
[19] Chang, C.-I., Du, Y., Wang, J., Guo, S.-M., Thouin, P.D.,”Survey and comparative analysis of entropy and relative entropy thresholding techniques”,
IEEE Proceedings Vision,Image and Signal Processing,
[20] Om Prakash Verma et al, “A Simple Single Seeded Region Growing Algorithm for Color Image Segmentation using Adaptive Thresholding”. 2011
International Conference on Communication Systems and Network Technologies.
[21] Savneet Dhaliwal, Abhilasha Jain “A Survey on Seeded Region Growing based Segmentation Algorithms” International Journal of Computer
Science and Management Research. Vol 2 Issue 6 June 2013 ISSN 2278-733X.
[22] Manisha Sharma, Vandana Chouhan “Objective Evaluation Parameters of Image Segmentation Algorithms”. International Journal of Engineering
and Advanced Technology (IJEAT) ISSN: 2249 – 8958, Volume-2, Issue-2, December 2012.
[23] B.Sathya, R.Manavalan “ Image Segmentation by Clustering Methods:Performance Analysis” International Journal of Computer Applications (0975
– 8887) Volume 29– No.11, September 2011
[24] B.Sowmya, B.Sheelarani “ Colour Image Segmentation Using Soft Computing Techniques”. International Journal of Soft Computing Applications
ISSN: 1453-2277 Issue 4 (2009), pp.69-80.
[25] A Paper on “A Seeded region growing algorithm:Review” authored by Shilpa Dantulwar and Prof.R.K.krishna has been published and presented
in International Conference on Recent Trends and innovations in Engineering &Technology-2013, ICRTIET 2013” Nagpur, Maharashtra , India in
association with Northern Illinois University.
[26] A Paper on “Optimization of color image segmentation using Seeded region growing algorithm:” authored by Shilpa Dantulwar and
Prof.R.K.krishna has been published in National conference at NCDRM 2014 organized by R.C.E.R.T. Chandrapur.
.

PERFORMANCE ANALYSIS USING SINGLE SEEDED REGION GROWING ALGORITHM

Recommended

More Related Content

What's hot (19)

Viewers also liked (11)

Similar to PERFORMANCE ANALYSIS USING SINGLE SEEDED REGION GROWING ALGORITHM (20)

More from AM Publications (20)

Recently uploaded (20)

PERFORMANCE ANALYSIS USING SINGLE SEEDED REGION GROWING ALGORITHM