# Dictionary:Hilbert-transform technique

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Other languages:
English • ‎español

A technique for determining the phase of a minimum-phase function from its power spectrum, used in computing a deconvolution operator. Given the power spectrum P(f) and that the wavelet is minimum phase, the wavelet's frequency-domain representation W(f) is

$W(f)=A(f)e^{i\gamma (f)}=\vert P(f)\vert ^{1/2}e^{i\gamma (f)}$ .

The amplitude A(f) is the square root of the power spectrum. Taking the logarithm of both sides splits the function into real and imaginary parts:

$\ln \left[W(f)\right]=(1/2)\ln \left[P(f)\right]+i\gamma (f)$ .

To be minimum phase, the function must be analytic in the lower half-plane. Then the Hilbert transform can be used to find the phase γ(f) from lnP(f)/2:

$\gamma (f)=(1/2)\ln \left[P(f)\right]*{\frac {1}{\pi f}}$ .

Since the amplitude and phase are known, the Fourier transform can be computed and the time-domain expression for W(t) determined. .