18.12.1.2 Algorithms (Continuous Wavelet Transform)


Continuous Wavelet Transform

This function computes the real continuous wavelet coefficient for each given scale presented in the Scale vector and each position b from 1 to n, where n is the size of the input signal.

Let x(t) be the input signal and ψ be the chosen wavelet function, the continuous wavelet coefficient of x(t) at scale a and position b is:

C_{a,b}  = \int_R {x(t)\frac{1}{{\sqrt a }}\psi^* } (\frac{{t - b}}{a})dt

The computation is implemented with a NAG function: nag_cwt_real(). It does not compute the coefficients with the definition of CWT. Instead, the integrals of the scaled, shifted wavelet function are approximated and the convolution is then computed.

Origin Wavelet Types

  • Morlet
    The Morlet wavelet:
    \psi (x) = \pi ^{ - 1/4} \cos (kx)e^{ - x^2 /2}
    where k is the wave number.
  • DGauss
    The Derivative of a Gaussian wavelet, which is the pth derivative of the Gaussian function:
    \psi (x) = (\frac{2}{\pi })^{ - 1/4} e^{ - x^2 }
    where p is the derivative order.
  • MexHat
    The Mexican Hat Wavelet:
    \psi (x) = \frac{2}{{\sqrt 3 }}\pi ^{ - 1/4} (1 - x^2 )e^{ - x^2 /2}

Convert Scale to Pseudo Frequency

For a given wavelet, you can map a scale and convert to pseudo-frequency by ways below:

F_a=\frac{F_c}{s\cdot \Delta }

In this formula:

  • s is a scale.
  • \Delta is the sampling period.
  • F_c is the center frequency of a wavelet in Hz.
  • F_a is the pseudo-frequency corresponding to the scale a, in Hz.

The F_c is the frequency contributes most to the variability of the wavelet, and it can be derived from maximizing the FFT of the wavelet modulus.

CWTFreq.png