Shift-invariance of short-time Fourier transform in fractional Fourier domains (Q1017567): Difference between revisions

From MaRDI portal
Set OpenAlex properties.
ReferenceBot (talk | contribs)
Changed an Item
 
Property / cites work
 
Property / cites work: Fractional Fourier domains / rank
 
Normal rank
Property / cites work
 
Property / cites work: The Fractional Order Fourier Transform and its Application to Quantum Mechanics / rank
 
Normal rank
Property / cites work
 
Property / cites work: Fast computation of the ambiguity function and the Wigner distribution on arbitrary line segments / rank
 
Normal rank
Property / cites work
 
Property / cites work: Short-time fourier transform: two fundamental properties and an optimal implementation / rank
 
Normal rank
Property / cites work
 
Property / cites work: The fractional fourier transform and the wigner distribution / rank
 
Normal rank
Property / cites work
 
Property / cites work: Linear operator theory in engineering and science. Repr. of the 1971 orig., publ. by Holt, Rinehart \& Winston, Inc. / rank
 
Normal rank

Latest revision as of 14:07, 1 July 2024

scientific article
Language Label Description Also known as
English
Shift-invariance of short-time Fourier transform in fractional Fourier domains
scientific article

    Statements

    Shift-invariance of short-time Fourier transform in fractional Fourier domains (English)
    0 references
    0 references
    12 May 2009
    0 references
    Let \(T\) be a linear time-frequency distribution with kernel \(K\) acting on a function \(x(t)\) evaluated at a time \(t\) and frequency \(f\) as follows \[ T(x) (t,f) = \int K(t, f, \tau) \, x(\tau)\, d\tau. \] For example, the short-time Fourier transform with window \(w\) is defined as \[ \text{{STFT}} (x) (t,f) = \int x(\tau) \overline{w}(t-\tau) \text{{e}}^{-2\pi i f \tau}\, d\tau. \] The author considers a certain covariance property of short-time Fourier transforms with respect to temporal shifts and fractional Fourier transforms. The latter can be defined as \[ \mathcal{F}_a(x) (t) = {\text{{e}}^{-\pi i \text{{sgn}}(a)/4 +\phi/2}\over \sqrt{|\sin\phi|}} \int \text{{e}}^{\pi i (t^2 \cot \phi -2t\tau csc \phi +\tau^2 \cot\phi)} \, x(\tau) \, d\tau \] where \(a\in [-2,2]\) denotes the order of the fractional Fourier transform. These transforms compose in the sense that \(\mathcal{F}_a\mathcal{F}_b=\mathcal{F}_{a+b}\) and \(\mathcal{F}_1\) is the usual Fourier transform. Let \(S_\rho (x)(t) = x(t-\rho)\) denote the usual shift operator and for \(\phi=a\pi/2\) let \(S_{\phi,\rho}\) denote the operator \(S_{\phi,\rho}=\mathcal{F}_{-a} S_\rho \mathcal{F}_a\). The author proves that any linear time-frequency distribution operator \(T\) that is ``magnitude-wise'' shift-covariant in the sense that \[ |T\circ S_{\phi,\rho} (x)(t,f)|=|T(x)(t-\rho \cos\phi,f-\rho\sin\phi)| \] must be a short-time Fourier transform with a suitable window function \(w\). The author illustrates this covariance property with some numerical examples.
    0 references
    fractional Fourier transform
    0 references
    time-frequency distribution
    0 references
    short-time Fourier transform
    0 references
    shift invariant
    0 references

    Identifiers