Feature sets for nonstationary signals derived from moments of the singular value decomposition of Cohen-Posch (positive time-frequency) distributions

This article presents a new method for determining the principal features of a nonstationary time series process based on the singular value decomposition (SVD) of the Cohen-Posch (1985) positive time-frequency distribution. This new method uses density functions derived from the SVD singular vectors to generate moments that are associated with the principal features of the nonstationary process. Since the SVD singular vectors are orthonormal, the vectors whose elements are composed of the squared elements of the SVD vectors are discrete density functions. Moments generated from these density functions are the principal features of the nonstationary time series process. The main reason for determining features of a time series process is to characterize it by a few simple descriptors.

[1]  Lane M. D. Owsley,et al.  Self-organizing feature maps and hidden Markov models for machine-tool monitoring , 1997, IEEE Trans. Signal Process..

[2]  Charles L. Lawson,et al.  Solving least squares problems , 1976, Classics in applied mathematics.

[3]  W. J. Williams,et al.  An algorithm for positive time-frequency distributions , 1996, Proceedings of Third International Symposium on Time-Frequency and Time-Scale Analysis (TFTS-96).

[4]  Patrick J. Loughlin,et al.  Time-frequency-based classification , 1996, Optics & Photonics.

[5]  G. Eichmann,et al.  An expansion of Wigner distribution and its applications , 1985, ICASSP '85. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[6]  Dale Groutage,et al.  A fast algorithm for computing minimum cross-entropy positive time-frequency distributions , 1997, IEEE Trans. Signal Process..

[7]  F. Grund.  Forsythe, G. E. / Malcolm, M. A. / Moler, C. B., Computer Methods for Mathematical Computations. Englewood Cliffs, New Jersey 07632. Prentice Hall, Inc., 1977. XI, 259 S , 1979 .

[8]  A. Laub,et al.  The singular value decomposition: Its computation and some applications , 1980 .

[9]  Michael A. Malcolm,et al.  Computer methods for mathematical computations , 1977 .

[10]  L. Cohen,et al.  Time-frequency distributions-a review , 1989, Proc. IEEE.

[11]  N. M. Marinovic,et al.  Feature Extraction And Pattern Classification In Space - Spatial Frequency Domain , 1985, Other Conferences.

[12]  Tzu-Hsien Sang,et al.  A new signal adaptive approach to positive time-frequency distributions with suppressed interference terms , 1998, Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP '98 (Cat. No.98CH36181).

[13]  Patrick J. Loughlin,et al.  Non-stationary signal classification using the joint moments of time-frequency distributions , 1998, Pattern Recognit..

[14]  L. Atlas,et al.  Automatic feature-finding for time-frequency distributions , 1996, Proceedings of Third International Symposium on Time-Frequency and Time-Scale Analysis (TFTS-96).

[15]  Les E. Atlas,et al.  Construction of positive time-frequency distributions , 1994, IEEE Trans. Signal Process..

[16]  L. Cohen.  Generalized Phase-Space Distribution Functions , 1966 .

[17]  Patrick J. Loughlin,et al.  An information-theoretic approach to positive time-frequency distributions , 1992, [Proceedings] ICASSP-92: 1992 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[18]  Leon Cohen,et al.  Positive quantum joint distributions , 1980 .

[19]  Leon Cohen,et al.  Positive time-frequency distribution functions , 1985, IEEE Trans. Acoust. Speech Signal Process..

[20]  Ben Pinkowski.  Principal component analysis of speech spectrogram images , 1997, Pattern Recognit..