Unsupervised filtering of Munsell spectra

  • M. Hauta-Kasarill
  • W. Wang
  • S. Toyooka
  • J. Parkkinen
  • R. Lenz
Poster Session I
Part of the Lecture Notes in Computer Science book series (LNCS, volume 1351)


We present a new method for producing color filters with positive coefficients to represent color reflectance spectra. The subspace method which is based on the KL-expansion can be used to define a basis to describe the spectral data accurately. However, due the orthogonality of the eigenvectors, the corresponding color filters usually contain negative coefficients and cannot be used in optical components directly. Our method finds the set of vectors which span a very similar color space as the subspace method does. These color filters contain only positive coefficients and can be directly used in optical implementations. We used an unsupervised competitive neural network (Instar) to find a set of positive color filters. The experiments with the Munsell spectra show that the filters produced by the neural network span a color space very similar to the color space spanned by the eigenvectors of the subspace method.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    P.K. Kaiser and R.M. Boynton, Human Color Vision, Second Edition, Optical Society of America, Washington DC, 1996.Google Scholar
  2. 2.
    G. Wyszecki and W.S. Stiles, Color science: concepts and methods, quantitative data and formulae, Wiley, New York, 1982.Google Scholar
  3. 3.
    M.S. Drew and B.V. Font, Natural Metamers, CVGIP: Image Understanding 56, 1992, 139–151.Google Scholar
  4. 4.
    J.P.S. Parkkinen, J. Hallikainen, and T. Jaaskelainen, Characteristic spectra of Munsell colors, J. Opt. Soc. Am. A. 6, 1989, 318–322.Google Scholar
  5. 5.
    T. Jaaskelainen, J. Parkkinen, and S. Toyooka, Vector-subspace model for color representation, J. Opt. Soc. Am. A. 7, 1990, 725–730.Google Scholar
  6. 6.
    Munsell Book of Color, Matte Finish Collection, Munsell Color, Baltimore, USA, (1976).Google Scholar
  7. 7.
    S. Usui, S. Nakauchi, and M. Nakano, Reconstruction of Munsell color space by a five-layer neural network, J. Opt. Soc. Am. A. 9, 1992, 516–520.Google Scholar
  8. 8.
    T. Jaaskelainen, S. Toyooka, S. Izawa, and H. Kadono, Color classification by vector subspace method and its optical implementation using liquid crystal spatial light modulator, Optics Communications 89, 1992, 23–29.Google Scholar
  9. 9.
    N. Hayasaka, S. Toyooka, and T. Jaaskelainen, Iterative feedback method to make a spatial filter on a liquid crystal spatial light modulator for 2D spectroscopic pattern recognition, Optics Communications 119, 1995, 643–651.Google Scholar
  10. 10.
    R. Lenz, M. Osterberg, J. Hiltunen, T. Jaaskelainen, and J. Parkkinen, Unsupervised filtering of color spectra, J. Opt. Soc. Am. A. 13, 1996, 1315–1324.Google Scholar
  11. 11.
    M. Osterberg, Quality Functions for Parallel Selective Principal Component analysis, Ph.D. dissertation, Linköping University, Linköping, Sweden, 1994.Google Scholar
  12. 12.
    E. Oja, Subspace Methods of Pattern Recognition, Research Studies Press, Letchworth, England, 1983.Google Scholar
  13. 13.
    S. Grossberg, Studies of the Mind and Brain, Reidel Press, Drodrecht, Holland, 1982.Google Scholar
  14. 14.
    T. Kohonen, The Self-Organizing Maps, Springer-Verlag, Berlin Heidelberg, 1995.Google Scholar
  15. 15.
    S. Haykin, Neural Networks, Macmillan College Publishing Company, New York, 1994.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • M. Hauta-Kasarill
    • 1
  • W. Wang
    • 1
  • S. Toyooka
    • 1
  • J. Parkkinen
    • 2
  • R. Lenz
    • 3
  1. 1.Department of Environmental Science and Human Engineering, Graduate School of Science and EngineeringSaitama UniversityUrawa, SaitamaJapan
  2. 2.Department of Information TechnologyLappeenranta University of TechnologyLappeenrantaFinland
  3. 3.Image Processing Laboratory, Department of Electrical EngineeringLinköping UniversityLinköpingSweden

Personalised recommendations