Advertisement

Orthogonal Nonnegative Matrix Factorization for Blind Image Separation

  • Andri Mirzal
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8237)

Abstract

This paper describes an application of orthogonal nonnegative matrix factorization (NMF) algorithm in blind image separation (BIS) problem. The algorithm itself has been presented in our previous work as an attempt to provide a simple and convergent algorithm for orthogonal NMF, a type of NMF proposed to improve clustering capability of the standard NMF. When we changed the application domain of the algorithm to the BIS problem, surprisingly good results were obtained; the reconstructed images were more similar to the original ones and pleasant to view compared to the results produced by other NMF algorithms. Good results were also obtained when another dataset that consists of unrelated images was used. This practical use along with its convergence guarantee and implementation simplicity demonstrate the benefits of our algorithm.

Keywords

nonnegative matrix factorization convergent algorithm blind image separation orthogonality constraint 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Ding, C., Li, T., Peng, W., Park, H.: Orthogonal nonnegative matrix tri-factorizations for clustering. In: 12th ACM SIGKDD Intl Conf. on Knowledge Discovery and Data Mining, pp. 126–135 (2006)Google Scholar
  2. 2.
    Lee, D., Seung, H.: Learning the parts of objects by non-negative matrix factorization. Nature 401, 788–791 (1999)CrossRefGoogle Scholar
  3. 3.
    Lee, D., Seung, H.: Algorithms for non-negative matrix factorization. In: Proc. Advances in Neural Processing Information Systems, pp. 556–562 (2000)Google Scholar
  4. 4.
    Mirzal, A.: A convergent algorithm for orthogonal nonnegative matrix factorization. Submitted to J. Computational and Applied MathematicsGoogle Scholar
  5. 5.
    Mirzal, A.: Nonnegative matrix factorizations for clustering and LSI: Theory and programming. LAP LAMBERT Academic Publishing, Germany (2011)Google Scholar
  6. 6.
    Lin, C.: On the convergence of multiplicative update algorithms for nonnegative matrix factorization. IEEE Transactions on Neural Networks 18(6), 1589–1596 (2007)CrossRefGoogle Scholar
  7. 7.
    Pauca, V., Piper, J., Plemmons, R.: Nonnegative matrix factorization for spectral data analysis. Linear Algebra and Its Applications 416(1), 29–47 (2006)MathSciNetCrossRefzbMATHGoogle Scholar
  8. 8.
    Cichocki, A., Amari, S.-i., Zdunek, R., Kompass, R., Hori, G., He, Z.: Extended smart algorithms for non-negative matrix factorization. In: Rutkowski, L., Tadeusiewicz, R., Zadeh, L.A., Żurada, J.M. (eds.) ICAISC 2006. LNCS (LNAI), vol. 4029, pp. 548–562. Springer, Heidelberg (2006)CrossRefGoogle Scholar
  9. 9.
    Kim, J., Park, H.: Toward faster nonnegative matrix factorization: A new algorithm and comparisons. In: 8th IEEE Intl Conf. on Data Mining, pp. 353–362 (2008)Google Scholar

Copyright information

© Springer International Publishing Switzerland 2013

Authors and Affiliations

  • Andri Mirzal
    • 1
  1. 1.Faculty of Computing N28-439-03Universiti Teknologi MalaysiaJohor BahruMalaysia

Personalised recommendations