I have the following (experimental) image, with its correponding raw data in .csv format here.


The pattern represents the scattering of a laser field off a point defect located in the center. One can clearly see a V-like shape of the scattered waves (here drawn in green). Unfortunately, there is also quite a lot of back-scattering going on, resulting in the very closely spaced parabolic waves running in the opposite direction.

I would like to find an fft-based method for filtering out all the secondary waves, leaving only the V-shaped ones.

  • 1
    $\begingroup$ There is no Fourier method that does what you want. Simply because the Fourier analysis of a static image does not (and cannot!) distinguish between waves travelling to the left and to the right. Since your scattering process seems to be elastic, the wavelength/spatial frequency is identical everywhere and the signal you want has the same frequency as the one you would like to discard. If you want to enhance your image you'll have to provide a sequence of consecutive frames of your scattering process. $\endgroup$ – Jazzmaniac Jul 1 '14 at 22:45
  • $\begingroup$ but the V-shaped feature is clearly not symmetrical around the defect, i.e. it appears on only one side. i dont care so much if i get rid of the small wavelength scattering in both directions, as long as i keep the large features. one of my goals is to extract the angle of the V $\endgroup$ – Andrei Berceanu Jul 2 '14 at 7:19
  • 1
    $\begingroup$ That's all fine, but has nothing to do with Fourier methods. Also I don't quite see why you can't read off the angle right now. $\endgroup$ – Jazzmaniac Jul 2 '14 at 7:45
  • $\begingroup$ it has to do with fourier methods, because it involves filtering out waves of a certain frequency $\endgroup$ – Andrei Berceanu Jul 2 '14 at 13:42
  • 1
    $\begingroup$ No, the frequency is not sufficiently different for the waves you want to remove. And again, what speaks against reading the angle from the picture as it is? $\endgroup$ – Jazzmaniac Jul 2 '14 at 14:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.