Positive time-frequency distributions via maximum entropy deconvolution of the evolutionary spectrum

The relationship between Priestley's definition of the evolutionary densities (TFDs) is explored, and a synthesis method is presented. As defined by Priestley, the ES is not a member of the Cohen-Posch class of TFDs. However, it is shown that by choosing a unit-energy normalization for the envelope function of Priestley's formulation, the energetic ES thus obtained is a member of the Cohen-Posch class of TFDs; this normalization differs from that chosen by Priestley. A method is then presented to obtain an estimate of the energetic ES. This method employs maximum entropy deconvolution of the spectrogram, which is itself a blurred version of the ES. Because the energetic ES is everywhere nonnegative and yields the correct marginal densities, it is a legitimate, joint time-frequency energy density of the signal, unlike the Wigner and other bilinear distributions that go negative.<<ETX>>

[1]  I. Csiszár.  Why least squares and maximum entropy? An axiomatic approach to inference for linear inverse problems , 1991 .

[2]  Langford B. White,et al.  Signal synthesis from Cohen's class of bilinear time-frequency signal representations using convex projections , 1991, IEEE International Conference on Acoustics, Speech, and Signal Processing.

[3]  Joseph A. O'Sullivan,et al.  Deblurring subject to nonnegativity constraints , 1992, IEEE Trans. Signal Process..

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

[5]  S. Gull,et al.  Image reconstruction from incomplete and noisy data , 1978, Nature.

[6]  I. Csiszár.  Why least squares and maximum entropy? An axiomatic approach to inverse problems , 1990 .

[7]  D. B. Preston.  Spectral Analysis and Time Series , 1983 .

[8]  Richard P. Bocker,et al.  Fast image reconstruction based on the regularized pseudoinverse of the imaging operator , 1992, Optics & Photonics.

[9]  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.

[10]  R. Harrison,et al.  The relationship between Wigner-Ville and evolutionary spectra for frequency modulated random processes , 1986, ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[11]  Nicolas S. Tzannes,et al.  An efficient algorithm for implementing the relative entropy method , 1988, ICASSP-88., International Conference on Acoustics, Speech, and Signal Processing.

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

[13]  Robert L. Burr.  Iterative convex I-projection algorithms for maximum entropy and minimum cross-entropy computations , 1989, IEEE Trans. Inf. Theory.

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