0

First - I need to plot this radially averaged spectrum 1-D, against the wavenumber |k|. But early I calculate the |k| using two matrix, so I have a matrix that represent |k|. So I have to radial average this matrix too?

Second - If I remove a mean value from my data before using the FFT2 function, I expected to see only a difference in the amplitude of the values, but actually I am noting a difference in the shape of the radially averaged spectrum.

I am trying to use this to analyse magnetic field data. Thanks for any help!!

Wanderson Ferreira
  • 1
  • 1
  • You might want to post this in the physics (or math?) stackexchange site. I am not exactly sure about the first point, but a little more info might help, for instance is |k| not *only* a function of the radius? – Buck Thorn Aug 05 '13 at 15:53
  • A more general question is: what kind of symmetry are you expecting (or observe) in your image? Is the operation that gives you |k| generating a matrix with that same symmetry? – Buck Thorn Aug 05 '13 at 16:00

1 Answers1

0

The explanation for your second question follows from the additive property of the FFT:

FFT(c) = FFT(a+b) = FFT(a) + FFT(b)

If b is an offset - namely the mean value - which you remove from signal c, then the spectrum of c-b will equal your original spectrum minus the spectrum of a constant, b, but the FFT of a constant results in sinc wiggles! So by removing the mean value of the signal you are removing a sinc glitch at zero frequency.

Buck Thorn
  • 5,024
  • 2
  • 17
  • 27