MATLAB: Wavelet

wavelet

For octave wavelet coefficients feature extraction take all the 8 wavelet coefficients or take only for 8th coefficient?

Best Answer

Hi Jaffino, the term "octave" as applied to the DWT is just referring to the dyadic scales. So you take all the coefficients at that scale. For example:

x = randn(1024,1);
[C,L] = wavedec(x,3,'db4');
d3 = detcoef(C,L,3);

d3 is the vector of coefficients at scale 3 (3rd octave). I think octave is a bit outdated now in the wavelet literature, better to refer to them as just scales.

Related Solutions

MATLAB: Undecimated wavelet transform

Hi Kalai, you can use either swt() or ndwt() (for a 1-D signal) to do the nondecimated wavelet transform.

Wayne

MATLAB: How to get the scales used with the “Dyadic default” option in the Wavelet Toolbox

The function "cwtft" returns a structure that among other things contains the scales used for the continuous wavelet transform. For this function the default wavelet type is Morlet which has a default value of 'scales' corresponding to the scales used with the "Dyadic default" option in the GUI. You can get these scales by passing "cwtft" any signal and reading the "scales" parameter of the structure. The following code loads a set of data shipped with MATLAB, passes it to "cwtft", and then stores the scales.

>> load kobe
>> out = cwtft(kobe);
>> dyadicDefault = out.scales

The documentation for "cwtft" also has the exact functions used for the "Default dyadic" scales under More About->Morlet Wavelet.

Best Answer

Related Solutions

MATLAB: Undecimated wavelet transform

MATLAB: How to get the scales used with the “Dyadic default” option in the Wavelet Toolbox

Related Question