Skip to main content

A SOM-based Chan–Vese model for unsupervised image segmentation

Abstract

Active Contour Models (ACMs) constitute an efficient energy-based image segmentation framework. They usually deal with the segmentation problem as an optimization problem, formulated in terms of a suitable functional, constructed in such a way that its minimum is achieved in correspondence with a contour that is a close approximation of the actual object boundary. However, for existing ACMs, handling images that contain objects characterized by many different intensities still represents a challenge. In this paper, we propose a novel ACM that combines—in a global and unsupervised way—the advantages of the Self-Organizing Map (SOM) within the level set framework of a state-of-the-art unsupervised global ACM, the Chan–Vese (C–V) model. We term our proposed model SOM-based Chan–Vese (SOMCV) active contour model. It works by explicitly integrating the global information coming from the weights (prototypes) of the neurons in a trained SOM to help choosing whether to shrink or expand the current contour during the optimization process, which is performed in an iterative way. The proposed model can handle images that contain objects characterized by complex intensity distributions, and is at the same time robust to the additive noise. Experimental results show the high accuracy of the segmentation results obtained by the SOMCV model on several synthetic and real images, when compared to the Chan–Vese model and other image segmentation models.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13

Notes

  1. For both simplicity and uniformity of notation, in writing (4) and other PDEs, we do not show explicitly the arguments of the functions, which are already specified in other parts of the paper.

  2. In this paper, training pixels from one image are considered in the training session of the SOMCV and \(SOMCV_s\) models. Such an image is either identical or similar to the image presented in the testing session. In the first case, using even identical images for the training and testing sessions is not a limitation of the models: one of the reasons is that the training is unsupervised.

  3. Instead of sequential training, batch training (Vesanto et al. 2000) of the SOM may also be used for a faster convergence.

  4. This choice of the function \(h_{bn}(t)\) implies that, for fixed t, when \(\Vert r_{b}-r_{n}\Vert \) increases, \(h_{bn}(t)\) decreases to zero gradually to smooth out the effect of the BMU neuron on the weights of the neurons far from the BMU neuron itself, and when t increases, the influence of the BMU neuron becomes more and more localized.

  5. The developed code is available at http://mohammedabdelsamea.weebly.com.

  6. In the experiments presented in Figs. 3 and 8, also the choice \(\sigma =.5\) was considered, together with \(\sigma = 1.5\).

  7. Since in the paper we are interested in unsupervised ACMs, we have not compared the proposed SOMCV model to the SOM-based ACMs proposed in Abdelsamea et al. (2014) and Abdelsamea et al. (2015), which are supervised models.

References

  • Abdelsamea MM, Gnecco G, Gaber MM, Elyan E (2015) On the relationship between variational level set-based and som-based active contours. Comput Intell Neurosci 2015. doi:10.1155/2015/109029

  • Abdelsamea MM, Gnecco G, Gaber MM (2014) A concurrent SOM-based Chan-Vese model for image segmentation. In: Proceedings of the 10th Workshop on Self-Organizing Maps (WSOM 2014), Advances in Intelligent Systems and Computing, vol 295, Springer, pp 199–208

  • Abdelsamea MM, Gnecco G, Gaber MM (2014) A survey of SOM-based active contours for image segmentation. In: Proceedings of the 10th Workshop on Self-Organizing Maps (WSOM 2014), Advances in Intelligent Systems and Computing, vol 295, Springer, pp 293–302

  • Abdelsamea MM, Tsaftaris SA (2013) Active contour model driven by globally signed region pressure force. In: Proceedings of the 18th International Conference On Digital Signal Processing

  • Abdelsamea MM, Gnecco G, Gaber MM (2015) An efficient self-organizing active contour model for image segmentation. Neurocomputing, Part B 149:820–835

    Article  Google Scholar 

  • Abdelsamea MM, Gnecco G (2015) Robust local-global SOM-based ACM. Electron Lett 51(2):142–143

    Article  Google Scholar 

  • Alpert S, Galun M, Basri R, Brandt A (2007) Image segmentation by probabilistic bottom-up aggregation and cue integration. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp 1–8

  • Bastos CACM, Tsang IR, Vasconcelos GS, Cavalcanti GDC (2012) Pupil segmentation using pulling & pushing and BSOM neural network. In: Proceedings of the 2012 IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp 2359–2364

  • Chan TF, Sandberg LA (2000) Active contours without edges for vector-valued images. J Vis Commun Image Represent 11(2):130–141

    Article  Google Scholar 

  • Chan TF, Vese LA (2001) Active contours without edges. IEEE Trans Image Process 10(2):266–277

    Article  MATH  Google Scholar 

  • Chen S, Radke RJ (2009) Level set segmentation with both shape and intensity priors. In: Proceedings of the 12th IEEE International Conference on Computer Vision, pp 763–770

  • Dongxiang C, Weizhong C (2010) A hybrid clustering method for automatic medical image segmentation. J Comput Inf Syst 6(6):1983–1993

    Google Scholar 

  • Haykin S (2009) Neural Networks and Learning Machines, 3rd edn, Pearson Prentice Hall, Upper Saddle River, NJ

  • Kohonen T (2013) Essentials of the self-organizing map. Neural Netw 37:52–65

    Article  Google Scholar 

  • Leventon ME, Grimson WEL, Faugeras O (2002) Statistical shape influence in geodesic active contours. In: Proceedings of the 5th IEEE EMBS International Summer School on Biomedical Imaging, pp 316–323

  • Li C, Kao C, Gore J, Ding Z (2008) Minimization of region-scalable fitting energy for image segmentation. IEEE Trans Image Process 17(10):1940–1949

    MathSciNet  Article  Google Scholar 

  • Li C, Huang R, Ding Z, Gatenby J, Metaxas D, Gore J (2011) A level set method for image segmentation in the presence of intensity inhomogeneities with application to MRI. IEEE Trans Image Process 20(7):2007–2016

    MathSciNet  Article  Google Scholar 

  • Liao P-S, Chen T-S, Chung P-C (2001) A fast algorithm for multilevel thresholding. J Inf Sci Eng 17(5):713–727

    Google Scholar 

  • Mian AS (2008) Realtime visual tracking of aircrafts. In: Proceedings of the 2008 IEEE International Conference on Digital Image Computing: Techniques and Applications, pp 351–356

  • Mumford D, Shah J (1989) Optimal approximations by piecewise smooth functions and associated variational problems. Commun Pure Appl Math 42(5):577–685

    MathSciNet  Article  MATH  Google Scholar 

  • Otsu N (1979) A threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern 9(1):62–66

    MathSciNet  Article  Google Scholar 

  • Shah-Hosseini H, Safabakhsh R (2003) A TASOM-based algorithm for active contour modeling. Pattern Recognit Lett 24(910):1361–1373

    Article  MATH  Google Scholar 

  • Skakun S (2010) A neural network approach to flood mapping using satellite imagery. Comput Inf 29(6):1013–1024

    MATH  Google Scholar 

  • Talu MF (2013) ORACM: Online region-based active contour model. Expert Syst Appl 40(16):6233–6240

    Article  Google Scholar 

  • Tsai A, Yezzi A, Wells W, Tempany C, Tucker D, Fan A, Grimson WE, Willsky A (2003) A shape-based approach to the segmentation of medical imagery using level sets. IEEE Trans Med Imaging 22(2):137–154

    Article  Google Scholar 

  • Vantaram SR, Saber E (2012) Survey of contemporary trends in color image segmentation. J Electron Imaging 21(4):28 Article ID: 040901

  • Vasconcelos GS, Bastos CACM, Tsang IR, Cavalcanti GDC (2011) BSOM network for pupil segmentation. In: Proceedings of the 2011 IEEE International Joint Conference on Neural Networks (IJCNN), 2011, pp 2704–2709

  • Venkatesh YV, Raja SK, Ramya N (2004) A novel SOM-based approach for active contour modeling. In: Proceedings of the 2004 Conference on Intelligent Sensors, Sensor Networks and Information Processing, pp 229–234

  • Venkatesh YV, Raja SK, Ramya N (2006) Multiple contour extraction from graylevel images using an artificial neural network. IEEE Trans Image Process 15(4):892–899

    Article  Google Scholar 

  • Venkatesh YV, Rishikesh N (2000) Self-organizing neural networks based on spatial isomorphism for active contour modeling. Pattern Recognit 33(7):1239–1250

  • Vesanto J, Himberg J, Alhoniemi E, Parhankangas J (2000) Self-organizing map in MATLAB: the SOM toolbox. In: In Proceedings of the Matlab DSP Conference, pp 35–40

  • Xu C, Prince JL (1998) Snakes, shapes, and gradient vector flow. IEEE Trans Image Process 7(3):359–369

    MathSciNet  Article  MATH  Google Scholar 

  • Zang K, Xu S, Zhou W, Liu B (2009) Active contours based on image Laplacian fitting energy. Chin J Electron 18(2):281–284

    Google Scholar 

  • Zhang K, Song H, Zhang L (2010) Active contours driven by local image fitting energy. Pattern Recognit 43(4):1199–1206

    Article  MATH  Google Scholar 

  • Zhang K, Zhang L, Song H, Zhou W (2010) Active contours with selective local or global segmentation: a new formulation and level set method. Image Vis Comput 28(4):668–676

    Article  Google Scholar 

  • Zhang K, Zhang L, Song H, Zhang D (2013) Reinitialization-free level set evolution via reaction diffusion. IEEE Trans Image Process 22(1):258–271

    MathSciNet  Article  Google Scholar 

  • Zhao H-K, Chan T, Merriman B, Osher S (1996) A variational level set approach to multiphase motion. J Comput Phys 127(1):179–195

    MathSciNet  Article  MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Giorgio Gnecco.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Communicated by V. Loia.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Abdelsamea, M.M., Gnecco, G. & Medhat Gaber, M. A SOM-based Chan–Vese model for unsupervised image segmentation. Soft Comput 21, 2047–2067 (2017). https://doi.org/10.1007/s00500-015-1906-z

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00500-015-1906-z

Keywords

  • Global region-based segmentation
  • Variational level set method
  • Active contours
  • Chan–Vese model
  • Self-organizing map
  • Neural networks