Novel Circular-Shift Invariant Clustering

Dimitrios Charalampidis

Abstract

Several important pattern recognition applications are based on feature extraction and vector clustering. Directional patterns may be represented by rota- tion-variant directional vectors, formed from M features uniformly extracted in M directions. It is often required that pattern recognition algorithms are invariant under pattern rotation or, equivalently, invariant under circular shifts of such directional vectors. This paper introduces a K-means based algorithm (Circular K-means) to cluster vectors in a circular-shift invariant manner. Thus, the algorithm is appropri- ate for rotation invariant pattern recognition applications. An efficient Fourier do- main implementation of the proposed technique is presented to reduce computa- tional complexity. An index-based approach is proposed to estimate the correct number of clusters in the dataset. Experiments illustrate the superiority of CK- means for clustering directional vectors, compared to the alternative approach that uses the original K-means and rotation-invariant vectors transformed from rotation- variant ones.

References

  1. Jain A.K. and Dubes R.C.: Algorithms for Clustering. Englewood Cliffs, N.J. Prentice Hall, (1988)
  2. Duda R.O. and Hart P.E.: Pattern Classification and Scene Analysis. New York, Wiley, (1973)
  3. Bezdek J.: Pattern Recognition with Fuzzy Objective Function Algorithms. New York, Plenum, (1981)
  4. Hartigan J.: Clustering Algorithms. New York, Wiley, (1975)
  5. Tou J. and Gonzalez R.: Pattern Recognition Principles. Reading, Mass., Addison-Wesley, (1974)
  6. Ruspini E.: A New Approach to Clustering. Information Control, Vol. 15, No. 1, (1969) 22- 32
  7. Su M.C. and Chou C.-H.: A Modified Version of the K-Means Algorithm with a Distance Based on Cluster Symmetry. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 23, No. 6, June (2001)
  8. Arof H. and Deravi F.: Circular Neighborhood and 1-D DFT Features for Texture Classification and Segmentation,” IEE Proceedings, Vision Image and Signal Processing, Vol. 145, No. 3, (1998) 167-172
  9. Charalampidis D. and Kasparis T.: Wavelet-based Rotational Invariant Segmentation and Classification of Textural Images Using Directional Roughness Features. IEEE Transactions on Image Processing, Vol. 11, No. 8, (2002) 825-836
  10. Haley G.M. and Manjunath B.S.: Rotation-invariant Texture Classification Using a Complete Space-Frequency Model. IEEE Trans. on Image Processing, Vol. 8, No. 2, (1999) 255-269
  11. Cohen F.S., Fan Z., and Patel M.A.: Classification of Rotated and Scaled Textured Images Using Gaussian Markov Random Field Models. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 13, No. 2, (1991) 192-202
  12. Gray R.M.: Toeplitz and Circulant Matrices: A Review. Web Address: http://ee.stanford.edu /gray/toeplitz.pdf
  13. Charalampidis D., Kasparis T., and Jones L.: Multifractal and Intensity Measures for the Removal of Non-precipitation Echoes from Weather Radars. IEEE Transactions on Geoscience and Remote Sensing, Vol. 40, No. 5, (2002) 1121-1131
  14. Georgiopoulos M., Dagher I., Heileman G.L. and Bebis G.: Properties of learning of a Fuzzy ART Variant. Neural Networks, Vol. 12, No. 6, (1999) 837-850
Download


Paper Citation


in Harvard Style

Charalampidis D. (2005). Novel Circular-Shift Invariant Clustering . In Proceedings of the 5th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2005) ISBN 972-8865-28-7, pages 33-42. DOI: 10.5220/0002568500330042


in Bibtex Style

@conference{pris05,
author={Dimitrios Charalampidis},
title={Novel Circular-Shift Invariant Clustering},
booktitle={Proceedings of the 5th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2005)},
year={2005},
pages={33-42},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0002568500330042},
isbn={972-8865-28-7},
}


in EndNote Style

TY - CONF
JO - Proceedings of the 5th International Workshop on Pattern Recognition in Information Systems - Volume 1: PRIS, (ICEIS 2005)
TI - Novel Circular-Shift Invariant Clustering
SN - 972-8865-28-7
AU - Charalampidis D.
PY - 2005
SP - 33
EP - 42
DO - 10.5220/0002568500330042