Artículos

 

Grayscale Image Segmentation Based on Associative Memories

 

Segmentación de imágenes en escala de gris con base en memorias asociativas

 

Enrique Guzmán Ramírez¹, Ofelia M. C. Jiménez¹, Alejandro D. Pérez¹, and Oleksiy Pogrebnyak²

 

¹ Universidad Tecnológica de la Mixteca, Oaxaca, Mexico. E–mail: eguzman@mixteco.utm.mx

² Centro de Investigación en Computación, Instituto Politécnico Nacional, México D. F., Mexico. E–mail: olek@pollux.cic.ipn.mx

 

Article received on 11/12/2010.
Accepted 05/02/2011.

 

Abstract

In this paper, a grayscale image segmentation algorithm based on Extended Associative Memories (EAM) is proposed. The algorithm is divided into three phases. First, the uniform distribution of the image pixel values is determined by means of the histogram technique. The result of this phase is a set of regions (classes) where each one is grouped into a certain number of pixel values. Second, the EAM training phase is applied to the information obtained at the first phase. The result of the second phase is an associative network that contains the centroids group of each of the regions in which the image will be segmented. Finally, the centroid to which each pixel belongs is obtained using the EAM classification phase, and the image segmentation process is completed. A quantitative analysis and comparative performance for frequently–used image segmentation by the clustering method, the k–means, and the proposed algorithm when it uses prom and med operators are presented.

Keywords: Image segmentation, associative memories, clustering techniques.

 

Resumen

En este artículo, un algoritmo para segmentación de imágenes en tonos de gris con base en las Memorias Asociativas Extendidas (EAM) es propuesto. El algoritmo es dividido en tres fases, en la primer fase se determina una distribución uniforme de los valores de los pixeles de la imagen utilizando la técnica de histograma. El resultado de esta fase es un conjunto de regiones (clases) donde cada una agrupa un determinado número de valores de pixel. En la segunda fase se aplica el algoritmo de entrenamiento de las EAM sobre la información obtenida en la primera fase; el resultado de esta fase es una red asociativa que contiene los centroides de las regiones que serán usadas en la segmentación de la imagen. En la última fase, usando el algoritmo de clasificación de las EAM se obtiene el centroide al cual cada uno de los pixeles de la imagen pertenece y el proceso de segmentación es completado. En la sección de resultados se presentan un análisis cuantitativo y una comparativa de desempeño, utilizando imágenes estándares de prueba, entre nuestra propuesta y un algoritmo de segmentación, basado en técnicas de clasificación, frecuentemente utilizado, el algoritmo k–means.

Palabras clave: Segmentación de imágenes, memorias asociativas, técnicas de clasificación.

 

DESCARGAR ARTÍCULO EN FORMATO PDF

 

References

1. Barron, R. (2006). Memorias asociativas y redes neuronales morfológicas para la recuperación de patrones. Tesis de doctorado, Centro de Investigación en Computación, Instituto Politécnico Nacional, México, D.F.         [ Links ]

2. Beevi, S.Z., Sathik, M.M. & Senthamaraikannan K. (2010). A Robust Fuzzy Clustering Technique with Spatial Neighborhood Information for Effective Medical Image Segmentation. International Journal of Computer Science and Information Security, 7(3), 132–138.         [ Links ]

3. Coello, C.A., Lamont, G.B. & Veldhuizen, D.A.V. (2007). Evolutionary Algorithms for Solving Multi–Objective Problems (2^nd ed.). New York: Springer.         [ Links ]

4. Deb, K. (2001). Multi–Objective Optimization Using Evolutionary Algorithms. New York: John Wiley & Sons.         [ Links ]

5. Dunn J.C. (1973). A Fuzzy Relative of the ISODATA Process and Its Use in Detecting Compact Well–Separated Clusters, Journal of Cybernetics, 3(3), 32–57.         [ Links ]

6. Handl, J. & Knowles, J. (2004). Multiobjective clustering with automatic determination of the number of clusters (Technical Report TR–COMPSYSBIO–2004–02). Manchester, UK, University of Manchester Institute of Science and Technology.         [ Links ]

7. Handl, J. & Knowles, J. (2007). An evolutionary approach to multiobjective clustering. IEEE Transactions on Evolutionary Computation, 11(1), 56–76.         [ Links ]

8. Jain A.K. (2010). Data clustering: 50 years beyond K–means. Pattern Recognition Letters, 31(8), 651–666.         [ Links ]

9. Jiang, X., Marti, C., Irniger, C. & Bunke, H. (2005). Image Segmentation Evaluation by Techniques of Comparing Clusterings. Image Analysis and Processing ICIAP 2005. Lecture Notes in Computer Science, 3617, 344–351.         [ Links ]

10. Jiang, X., Marti, C., Irniger, C. & Bunke, H. (2006). Distance Measure for Image Segmentation Evaluation, EURASIP Journal on Applied Signal Processing. 2006, 1 –10.         [ Links ]

11. Ristic, D.M., Pavlovic, M. & Reljin, I. (2008). Image Segmentation Method Based on Self Organizing Maps and K–Means Algorithm. 9^th Symposium on Neural Networ/ Applications in Electrical Engineering, Belgrade, Serbia, 27–30.         [ Links ]

12. Samma, A.S.B. & Salam, R.A. (2009). Adaptation of K–Means Algorithm for Image Segmentation. World Academy of Science, Engineering and Technology 50, 58–62.         [ Links ]

13. Shirakawa, S. & Nagao, T. (2009). Evolutionary Image Segmentation Based on Multiobjective Clustering. IEEE Congress on Evolutionary Computation, Trondheim, Norway, 2466–2473.         [ Links ]

14. Sossa, H., Barrón, R. & Vázquez, R.A. (2004). Real–valued Patterns Classification based on Extended Associative Memory. Fifth Mexican International Conference on Computer Science. Colima, México. 213–219.         [ Links ]

15. Steinbuch, K. (1961). Die Lernmatrix. Biological Cybernetics, 1(1), 36–45.         [ Links ]

16. Studholme C., Hill, D.L.G. & Hawkes, D.J. (1999). An overlap invariant entropy measure of 3D medical image alignment. Pattern Recognition, 32(1), 71–86.         [ Links ]

17. Tatiraju, S. & Mehta, A. (2008). Image Segmentation using k–means clustering, EM and Normalized Cuts (Technical Report). University Of California, Irvine. http://www.ics.uci.edu/~dramanan/teaching/ics273a_winter08/projects/avim_report.pdf        [ Links ]

18. The Berkeley Segmentation Dataset and Benchmark. (s.f.). Retrieved from http://www.eecs.berkeley.edu/Research/Projects/CS/vision/bsds/        [ Links ]

19. Xue–xi, Z. & Yi–min, Y. (2008). Hybrid Intelligent Algorithms for Color Image Segmentation. Chinese Control and Decision Conference. Yantai, China. 264–268.         [ Links ]

20. Yang, Z., Chung, F.L. & Shitong, W. (2009). Robust fuzzy clustering based image segmentation, Applied Soft Computing Journal, 9(1), 80–84.         [ Links ]

21. Yanling, L. & Yi, S. (2008). Robust Image Segmentation Algorithm Using Fuzzy Clustering Based on Kernel–Induced Distance Measure. International Conference on Computer Science and Software Engineering 2008, Wuhan, China, 1, 1065–1068.         [ Links ]

22. Yanling, L. & Yi, S. (2010). Fuzzy c–means clustering based on spatial neighborhood information for image segmentation. Journal of Systems Engineering and Electronics, 21(2), 323–328.         [ Links ]