Discrimination by optimizing a local consistency criterion

  • A. Zighed
  • D. Tounissoux
  • J. P. Auray
  • C. Largeron
6. Information
Part of the Lecture Notes in Computer Science book series (LNCS, volume 521)


The paper we present here offers a method of pattern recognition which could be considered as a non-parametrical technique based on geometrical procedures. The ideas developed can be found in certain approaches for restoring pictures with sound and in supervised learning algorithms based on relaxation.

Our approach is based on three points:
  • The definition of a neighbourhood structure endowed with certain properties.

  • Finding a local consistency criterion which we will try to optimize with a view to relabeling

  • Adopting a labeling rule for anonymous individuals.

We will conclude by presenting a few experimental results.


Discrimination Relabeling Neighbourhood Geometrical approach local consistency relaxation 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

8. Bibliography

  1. [1]
    P. Devijver Selection of prototypes for nearest neighbour classification. Indian statistical institute golden jubilee Proc. Int. Conf. on advances in information sciences and technology.1982Google Scholar
  2. [2]
    P. DEVIJVER & M. DEKESEL, Computing multidimensional Delaunay tesselation Report R.464 Philips Research Laboratory. Brussels., 1982Google Scholar
  3. [3]
    R. Fages, M. Terrenoire, D. Tounissoux, A. Zighed Non supervised classification tools adapted for supervised classification. Nato ASI Series Vol. F30 Springer-Verlag.1985Google Scholar
  4. [4]
    O. D. Faugeras & M. Berthod. Improving consistency and reduction ambiguity in stochastic labelling: an optimization approach. IEEE Vol. PAMI-30, No4, July 1981Google Scholar
  5. [5]
    K. Fukunaga. Introduction to statistical pattern recognition. Academic Press. 1972.Google Scholar
  6. [6]
    R. GASTINEL, Analyse numérique linéaire Herman Paris 1966.Google Scholar
  7. [7]
    X. Guyon & J.F. Yao Analyse discriminante contextuelle. Actes 5o Jr. Anl. des données et informatique. Tomel p43–52. 1987Google Scholar
  8. [8]
    Hossam A. El Gindy, Godfried T. Toussaint Computing the relative neighbour decomposition of simple polygon. Computational morphology Ed. G.T. Toussaint North-Holland 1988.Google Scholar
  9. [9]
    R.A. Hummel & S.W. Zucker On the foundation of relaxation labeling IEEE VOL. PAMI-5, NO. 3, May 1983Google Scholar
  10. [10]
    J. Illingworth & J. Kittler Optimisation algorithms in probabilistic relaxation labelling Pattern recognition Theory and application. Nato series Vol. 30 Springer Verlag. 1987Google Scholar
  11. [11]
    M. Jambu, M.O. Lebeaux Classification automatique pour l'analyse des données Dunod 1978Google Scholar
  12. [12]
    A.K. Jain Advances in statistical pattern recognition Pattern recognition Theory and application. Nato series Vol. 30 Springer Verlag. 1987Google Scholar
  13. [13]
    J. Kittler Relaxation labelling Pattern recognition Theory and application. Nato series Vol. 30 Springer Verlag. 1987Google Scholar
  14. [14]
    M. Levy A new theoriktical approach to relaxation, application to edge detection. Actes IAPR Rome 1988 IEEEGoogle Scholar
  15. [15]
    J. Marichy, G. Buffet, A. Zighed, P. Laurent Early detection of septicemia in burnt patients, Actes 3rd INt. Conf. Sci. in Health Care, p 505–508 Munich 1984 Ed. Springer-Verlag.Google Scholar
  16. [16]
    F. P. Preparata & M. I. Shamos. Computational geometry: an introduction. Springer-verlag. 1988Google Scholar
  17. [17]
    M. Terrenoire, D. Tounissoux. Restauration d'image par optimisation d'un critère d'homogénéité locale. Proceding of workshop on syntactical and structural pattern recognition. Pont-à-Mousson. 1988Google Scholar
  18. [18]
    G. T. Toussaint. Computational geometry recent relevant to pattern recognition. Nato Asi Series Vol.F30 Springer-Verlag. 1987.Google Scholar
  19. [19]
    G. T. Toussaint. A Graph-theorical primal sketch. Computational morphology. E.d Toussaint, North-Holland 1988.Google Scholar
  20. [20]
    J. I. Toriwaki, S. Yokoi. Voronoi andrelated neighbors on digitized 2-dimensiona space with application of texture analysis. Computational morphology. E.d Toussaint, North-Holland 1988.Google Scholar
  21. [21]
    D. F. Watson. Computational the n-dimensional Delaunay tesselation with application to voronoi polytopes. The computer journal. Vol. 24 No2 1981.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1991

Authors and Affiliations

  • A. Zighed
    • 1
  • D. Tounissoux
    • 1
  • J. P. Auray
    • 1
  • C. Largeron
    • 1
  1. 1.URA 934 Bt 101University of Lyon IVilleurbanne

Personalised recommendations