Professional Documents
Culture Documents
Evolutionary Image Segmentation By Pixel Classification and the evolutionary Xie and Beni criterion Application to quality control
M. Merzougui1, A.EL Allaoui1, M. Nasri1, M. EL Hitmy2 and H. Ouariachi 1 1 LABO MATSI, ESTO, B.P 473, University of Mohammed I, OUJDA, MOROCCO.
2
Abstract
In this paper, we propose a segmentation method based on pixel classification and evolution strategies. Before segmentation, we start by determining the number of classes by means of an evolutionary Xie and Beni criterion. The proposed approach is validated on some synthetic and real images and, it shows to be very interesting as decision support in quality control. Keywords: Segmentation, evolutionary segmentation, Classification, evolutionary strategies, Segmentation by pixel classification, evolutionary Xie and Beni criterion.
1. Introduction
The segmentation is an essential stage in image processing. There are many consistent methods available today for image segmentation, among these, there is the segmentation based on pixels classification as a function of their grey level values [1][7][8][12]. Every pixel in the image holds an inherent relationship with the pixels in its surrounding. The information at a particular pixel may be in relation with the information over the whole or part of the image. We may be interested in the mean or the median values of the pixels grey level. The process of segmentation by pixel classification consists of three stages: [1] [7] [8] [12] [23]. Acquisition of data for each pixel in order to form the attribute vector Estimation of the number of classes with the evolutionary Xie and Beni criterion Pixel classification based on the acquired information Acquisition stage For each pixel, we compute two values, the mean value of the grey levels (MGL), and the difference between the MGL and the maximum value of the grey levels surrounding the particular pixel (DMGL). For this purpose, we use a square window centred at the particular pixel. Estimate the number of classes We use the evolutionary Xie and Beni criterion Classification stage Kmeans and evolutionary Kmeans are used for this classification purpose. The two algorithms use the information provided by the parameters MGL and DMGL, associated with each and individual pixel in order to classify the pixel with respect to centers that evolve at each iteration In Section 2 image segmentation with Kmeans algorithm is presented. Section 3 gives an introduction to evolution strategies approach and the proposed evolutionary Kmeans algorithm along with the image evolutionary segmentation. The estimation of the number of classes with the evolutionary Xie and Beni criterion is presented in section 4. In section 5, a validation of our approach is given; experimental results are obtained over some synthetic and real images. Finally we give a conclusion.
2. Kmeans classification
2.1. Descriptive elements
Consider a set of M objects {O1, O2,..., OM} characterized by N attributes, grouped in a line vector form
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8 V = (a1 a2 ... aN). Let Ri = (aij) 1jN be a line vector of RN where aij is the value of the attribute aj for the object Oi. Let mat_obs be a matrix of M lines (representing the objects Oi) and N columns (representing the attributes aj): (1) mat _ obs = aij 1i M
( )
1 j N
V is the attribute vector, Ri is the observation associated with Oi or the realization of the attribute vector V for this object, RN is the observations space [1] and mat_obs is the observation matrix associated with V. The ith line of mat_obs is the observation Ri. Ri belongs to a class CLs, s=1, , C.
J=
where
1 M C Ri g s 2 i =1 s =1
(2)
3. Evolution strategies
Evolutionary strategies (ES) are particular methods for optimizing functions. These techniques are based on the evolution of a population of solutions which under the action of some precise rules optimize a given behaviour, which initially has been formulated by a given specified function called fitness function [2][18]. An ES algorithm manipulates a population of constant size. This population is formed by candidate points called chromosomes. Each of the chromosomes represents the coding of a potential solution to the problem to be solved, it is formed by a set of elements called genes, these are reals. At each iteration, called generation, is created a new population from its predecessor by applying the genetic operators: selection and mutation. The mutation operator perturbs with a Gaussian disturbance the chromosomes of the population in order to generate a new population permitting to further optimize the fitness function. This procedure allows the algorithm to avoid the local optimums. The selection operator consists of constructing the population of the next generation. This generation is constituted by the pertinent individuals [2] [3][11]. Figure 1 illustrates the different operations to be performed in a standard ES algorithm [2][3] : Random generation of the initial population Fitness evaluation of each chromosome Repeat Select the parents Update the genes by mutation Select the next generation Fitness evaluation of each chromosome Until Satisfying the stop criterion
Figure 1: Standard SE algorithm.
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
(3) = g11..g1N g 21..g 2N ..g s1..g sN ..gC1..gCN The chr chromosome is a real line vector of dimension CN. The genes (gsj)1jN are the components of the gs center: (4) g s =(g sj )1 j N =(g s1g s2..g sj..g sN ) To avoid that the initial solutions be far away from the optimal solution, each chromosome chr of the initial population should satisfy the condition:
(5)
In the EKM algorithm, we discard any chromosome with a gene that does not satisfy this constraint. This gene, if any, is replaced by an other one which complies with the constraint [18] [23].
F (chr ) =
2 1 M C Ri g s M i =1 s =1
(6)
where chr* is the new chromosome obtained by a Gaussian perturbation of the old chromosome chr. N(0,1) is a Gaussian disturbance of mean value 0 and standard deviation value 1, is the strategic parameter. is high when the fitness value of chr is high. When the fitness value of chr is low, must take very low values in order to be not far away from the global optimum. We have been inspired from this approach to propose a new form of the mutation operator. The fact that we have proposed a new mutation operator is motivated by our interest to reach the global solution in a small computational time. Let chr be a chromosome of the population formed by the centers (gs)1sC. Let Ri CLs if Ri g s = min s '=1,C Ri g s ' , i.e. the class consisting of the Ri observations that are closest to the center gs. Let gs be the center of gravity of the CLs class
gs =
Ri Ri CLs ls
where ls =card(CLs )
(8)
The mutation operator which we propose in this work consists in generating, from the chr, the new chromosome chr* formed by the centers (g*s)1sC, as: g*s = gs + fm ( gs - gs) N(0,1) (9)
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
where fm is a multiplicative constant factor taken to be randomly chosen between 0.5 and 1. The new strategic parameter proposed = fm ( gs - gs ) is low when gs gets closer to gs and is high when gs is far from gs. The proposed has two advantages: - When chr is far from the global solution, chr is subjected to a strong Gaussian perturbation allowing chr to move more quickly in the research space and in the same time to avoid local solutions. - controls the Gaussian perturbation level. Indeed, as the chromosome chr gets closer to the global solution, the Gaussian perturbation level is reduced until becoming null at convergence. Generating children from parent chromosomes we have adopted the technique of selection by ordering. We have also used the elitist technique [10].
1+ls
where ls =card(CLs )
1.6. Compute for each chr of P its fitness value F(chr). Stage 2: Repeat 2.1. Order the chromosomes chr in P from the best to the poor ( in an increasing order of F). 2.2. Choose the best chromosomes chr. 2.3. For each chr of P attribute, the observations Ri to the corresponding classes:
1+ls
where ls =card(CLs )
(The population P obtained after the updating is the population of the next generation ) 2.8. Compute for each chr of P its fitness value F(chr). Until Nb_gen (generation number) ! maxgen
Figure 2: The proposed EKM algorithm.
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
Table 1 shows the classes of SYNTH1 along with the grey level values and the number of pixels in each class.
Table 1: Information on SYNTH1
Class 1 2 3
We considered for this test the attributes (MGL, DMGL) in a 3 * 3 window. We obtain for the Xie and Beni function the values shown in table2 and figure 4:
Table2: FXB for different values of C
C FXB
5 0.0978
6 0.6419
0.0153 0.1229
0.6
0.5
0.4
FXB
0.3 0.2 0.1 0 3
3.5
4.5
5.5
EKM
KM
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
10
15
20
25
30
Synth1 image segmentation results are summarised in table 3. Table 3 : rsults of segmentation for Synth1 image Initial image 12748 5736 119 description Background Triangle Line EKM 12748 5403 119 KM 12410 5855 0
The results show that the EKM clearly The The EKM detects the three objects of the image, background, triangle, and line. The KM was not able to detect the line. The number of misclassified pixels for both methods EKM and KM is respectively 333, 457. The corresponding error rate is:
EKM
333
18603
= 1.8%
KM
The two algorithms were run several times and we noticed that the EKM obtains each time the same result while the KM obtains different results. We can conclude that the EKM is the most stable; it outperforms the KM and obtains good results
We obtain for the Xie and Beni function the values shown in table4 and figure 8:
10
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
6
0.0564
7
0.2219
8
0.3907
Pad image segmentation with C=4 The segmentation results in this case are shown in figure 9 for three different running of the KM and EKM algorithms. The result sow that EKM result clearly detects the damage accruing on one of the tablets for the three running while the KM fails to detect the defect for each.
EKM KM
EKM
KM
EKM
KM
EKM
KM
Class number CL1 CL2 CL3 CL4 CL1 CL2 CL3 CL4
Trial 1
92 1197 1253 3541 973 1124 1140 2846
Trial 2
92 1197 1253 3541 973 1110 1153 2847
Trial 3
92 1197 1253 3541 973 1124 1140 2846
11
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
2500
2000
1500
1000
500
10
15
20
25
30
Figure 10 shows the convergence of EKM algorithm, convergence is achieved very quickly, in three iterations, table5 shows the results obtained by KM and EKM algorithms with details obtained on the classes for each running test. The EKM is more stable and has outperformed the KM algorithm.
7. Conclusion
Kmeans image segmentation shows some stability difficulties due to the initialisation problem of the KM. We have proposed the evolutionary Kmeans image segmentation in order to get around this difficulty. The proposed approach has been validated on synthetic and blister pad images. The experimental results obtained show the rapid convergence and the good performance of this approach. The instability problem has been eliminated. This approach may be used for problems of decision support in quality control.
8. Rfrences
[1]
Cocquerez T.P. et Phillip S. "Analyse d'images : Filtrage et segmentation". Editions MASSON, Paris, 1995.
[2] Hall, L. Q., zyurt, I. B. et Bezdek, J. C. " Clustering with a genetically optimized approach ". IEEE Trans. on Evolutionary Computation, Vol.3, N2, pp. 103-110, july 1999. [3] Presberger, T., Koch, M. " Comparison of evolutionary strategies and genetic algorithms for optimization of a fuzzy controller ". Proc. of EUFIT95, Aachen, Germany, august 1995. [4] Florence Huet et Sylvie Phlipp Etude de contours haute-chelle pour la segmentation et la fermeture de contours en prsence de zones textures et/ou bruites. Traitement du Signal 1998 Volume 15 - n 1 [5] Xie, X. and Beni, G. "A validity measure for fuzzy clustering". IEEE Trans. on Pattern Analysis and Machine Intell., Vol. 13, N 8, pp. 841-847, 1991. [6] Barra, V. : Fusion dImages 3D du Cerveau : Etude de Modles et Applications : Ph.D. Thesis, Universit d'Auvergne, Clermont-Ferrand (France), 2000. [7] Christophe Saint-Jean, Classification paramtrique robuste partiellement supervise en reconnaissance des formes, Thse de Doctorat, universit de La Rochelle - UFR Sciences Laboratoire dInformatique et dImagerie Industrielle 2001. [8] N.Vandenbroucke, L.Macaire, J.G. Postaire, Color image segmentation by pixel classification in an adapted hybrid color space. Application to soccer image analysis., Computer Vision and Image Understanding 90 (2), 190-216, (2003). [9] Zitouni Rafik, Segmentation Contextuelle des images IRM du cerveau humain (Application de la classification Semi Supervise), Thse dingniorat, universit de Stif, institut dinformatique 2004. [10] Nasri M. Contribution la classification de donnes par Approches Evolutionnistes : Simulation et Application aux images de textures. Thse de doctorat. Universit Mohammed premier Oujda 2004. [11]
Nasri M, M. EL Hitmy, H. Ouariachi and M.Barboucha. Optimization of a fuzzy classification by evolutionary strategies. In proceedings of SPIE Conf.,6 th international conference on quality control by artificial vision,vol.5132,pp. 220230,USA, 2003.Repulished as an SME Technical Paper by The society of manufacturingn .
12
International Journal of Computational Intelligence and Information Security, August 2011 Vol. 2, No. 8
[12] L. Macaire, N. Vandenbroucke, J.G Postaire Segmentation d'images par classification spatiocolorimtrique des pixels. Traitement du signal 2004 vol 21 N spcial L'image numrique couleur 423-437 [13] Z. Rafik1, B. Ihssen. Segmentation Contextuelle Des Images IRM Crbrales par Le FCM Semi Supervis. Premire Etape a lAide au Diagnostic. Journes Francophones dInformatique Mdicale, Lille 12-13 mai 2005 [14] Semchedine, M., Toumi, L., Moussaoui, A. : Nouvelle Approche de Classification Multimodale Hybride dImages IRM dans un SMA , Journes internationales sur linformatique graphique .Tbessa 2006 [15] Dou, W. : Segmentation d'images multi spectrales base sur la fusion dinformations : application aux images IRM. Thse doctorat de luniversit de Caen, Spcialit : Traitement du Signal et des Images, 2006 [16] Non-dominated Sorting Evolution Strategy-based K-means Clustering Algorithm for Accent Classification. S. Ullah, F. Karray, and J.M. Won. 978-1-4244-2175-6/08/2008 IEEE [17] Glorennec, P., Y. "Algorithmes dapprentissage pour systmes dinfrence floue". Colloque Neuromimtisme, Lyon, juin 1994. [18] H.Ouariachi, Classification non Supervise de donnes par les rseaux de neurones et par une approche volutionniste : application la segmentation dimages. Thse de doctorat. Universit Mohammed premier Oujda 2001. [19] A. EL Allaoui, M. Merzougui, M. Nasri, M. EL Hitmy and H. Ouariachi. Optimization of Unsupervised Classification by Evolutionary Strategies. IJCSNS International Journal of Computer Science and Network Security, ISSN: 1738-7906, Vol. 10 No. 6 pp. 325-332 June, 2010. [20] M. Merzougui, A. EL Allaoui, M. Nasri, M. EL Hitmy and H. Ouariachi. Unsupervised classification using evolutionary strategies approach and the Xie and Beni criterion. IJAST International Journal of Advanced Science and Technology, ISSN: 2005-4238, Vol. 19, pp 43-58 June, 2010. [21]
Bezdek, J. C. "Cluster validity with fuzzy sets". J. Cybernetics, Vol.3, N3, pp. 58-73, 1974.
[22] Karayiannis, N. B., Bezdek, J. C. et Hathaway, R. J. "Repairs to GLVQ : a new family of competitive learning schemes". IEEE Trans. on Neural Networks, Vol.7, N5, pp. 1062-1071, 1996. [23] A. EL Allaoui, M. Merzougui, M. Nasri, M. EL Hitmy and H. Ouariachi. Evolutionary image segmentation by pixel classification. Application to medical images. IJCIIS International Journal of Computational Intelligence and Information Security, ISSN: 1837-7823, March 2011, Vol. 2, No. 3, (p. 12-24).
9. Biography
Mohammed Merzougui works on image processing and pattern recognition in MATSI laboratory of the school of Technology, University of Oujda, Morocco. He has his master in the field of computer sciences at the same University. His main field of research interest is classification of texture images based on new methods (genetic and evolutionary approaches, neural networks, fuzzy logic). Ahmad EL ALLAOUI works on image processing and pattern recognition in MATSI laboratory of the school of Technology, University of Oujda, Morocco. He has his master in the field of computer sciences at University the Med Benabdellah at Fez. His main field of research interest is classification of texture based on new methods (genetic and evolutionary approaches, neural networks, fuzzy logic). Mbarek Nasri is a senior lecturer at the School of technology of Oujda University since 1996. He obtained his Ph.D degree in 2004 in the field of data Classification from the University of Oujda, department of physics. He obtained the habilitation degree in 2006 from the same university. His field of research interest is in image processing and its application to the quality control by artificial vision and medical imaging
13