Non stationary signals classification using time-frequency distributions

The paper deals with a comparison between different non-parametric classification methods of non stationary signals. The first ones, consider the time frequency representation (TFR) of the signal as the code itself and the decision is taken following the value of a dissimilarity index between the TFRs. In the other methods, the authors compute the instantaneous log-deviation between the (positive) TFR of the signal to be classified, and the TFR of each cluster. The classification results of each method (misclassification rate versus the cardinal of the learning set) are presented, and the influence of the choice of the TFR is studied.<<ETX>>

[1]  Richard O. Duda,et al.  Pattern classification and scene analysis , 1974, A Wiley-Interscience publication.

[2]  R. Altes.  Detection, estimation, and classification with spectrograms , 1980 .

[3]  S.M. Kay,et al.  Spectrum analysis—A modern perspective , 1981, Proceedings of the IEEE.

[4]  Boualem Boashash,et al.  Recognition of time-varying signals in the time-frequency domain by means of the Wigner distribution , 1984, ICASSP.

[5]  Boualem Boashash,et al.  Methods of signal classification using the images produced by the Wigner-Ville distribution , 1991, Pattern Recognit. Lett..

[6]  Eric Le Carpentier,et al.  An Optimal Approach for Random Signals Classification , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  F. Hlawatsch,et al.  Linear and quadratic time-frequency signal representations , 1992, IEEE Signal Processing Magazine.

[8]  Patrick Flandrin,et al.  Generalization of the reassignment method to all bilinear time-frequency and time-scale representations , 1994, Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing.