RECURSION IN SHORT-TIME SIGNAL ANALYSIS

The problem of finding a recursive structure for the evaluation of features through a ~running' window is investigated. A general closed form expression is found for features satisfying a direct or indirect recursion condition. It is shown that most of the commonly used features (mean value, energy, autocorrelation function, DFT, Z-transform, entropy, etc.) satisfy these analytic expressions. The recursive, step by step, feature evaluation method is compared with the conventional method where features are evaluated for positions of the observation window with a 50% overlap. These two methods are equivalent in computation time for features satisfying the direct recursion condition. However, there might be some loss of information when using the last approach. The use of indirect recursion is advantageous for small window sizes. The results are then generalised to bidimensional signal processing. Zusammenfassung. Es wird untersucht, wie man Parameter eines gleitenden Fensters rekursiv schfitzen kann. Es wird ein allgemeiner Ausdruck fiir den Fall angegeben, das die Sch~itzfunktion einer direkten oder indirekten Rekursionsbedingung geniigt. Dfesen Rekursionsbedingungen genfigen z.B. Mittelwert, Energie, die Autokorrelationsfunktion, die diskrete Fouriertransformation, die z-Transformation, die Entropie u.a.m. Die rekursiv gesch~itzten Parameter werden mit den iiblichen Sch~itzwerten verglichen, wenn die Fenster je zu 50% iiberlappen. Geniigt ein Parameter der direkten Rekursions- bedingung, sind beide Berechnungsmethoden beziiglich Rechenzeit ~iquivalent. Allerdings kann bei der letztgenannten Methode unter Umst(inden ein Teil der Information verloren gehen. Die indirekte Rekursion hat besonders bei kleinen Fenstern Vorteile. Die Ergebnisse werden auf den zweidimensionalen Fall verallgemeinert.

[1]  A. J. Jerri Correction to "The Shannon sampling theorem—Its various extensions and applications: A tutorial review" , 1979 .

[2]  B. Goldberg A continuous recursive DFT analyzer--The discrete coherent memory filter , 1980 .

[3]  P. Kumar,et al.  Theory and practice of recursive identification , 1985, IEEE Transactions on Automatic Control.

[4]  R. Bitmead On recursive discrete Fourier transformation , 1982 .

[5]  Benjamin Friedlander A recursive maximum likelihood algorithm for ARMA spectral estimation , 1982, IEEE Trans. Inf. Theory.

[6]  A. J. Jerri The Shannon sampling theorem—Its various extensions and applications: A tutorial review , 1977, Proceedings of the IEEE.

[7]  J. H. Halberstein Recursive, complex Fourier analysis for real-time applications , 1966 .

[8]  Thomas S. Huang,et al.  A fast two-dimensional median filtering algorithm , 1979 .

[9]  Gabriel E. Lowitz Extraction de caricatures et texture depuis un histogramme local , 1981 .

[10]  G. Bongiovanni,et al.  Procedures for computing the discrete Fourier transform on staggered blocks , 1976 .

[11]  Murat Kunt,et al.  Preprocessing of electrocardiograms by digital techniques , 1982 .

[12]  R.M. Haralick,et al.  Statistical and structural approaches to texture , 1979, Proceedings of the IEEE.

[13]  Bradley W. Dickinson Two recursive estimates of autoregressive models based on maximum likelihood , 1978 .

[14]  Robert M. Haralick,et al.  Textural Features for Image Classification , 1973, IEEE Trans. Syst. Man Cybern..

[15]  M.G. Bellanger,et al.  Digital processing of speech signals , 1980, Proceedings of the IEEE.