2

When I apply a Fourier transform to an image, I get huge complex values. What are the corresponding coefficients' values of high frequencies, and how about low frequencies? When I use the function fft2 in MATLAB, I get N x M values, for example: two coefficients: -2+3i and 2+2i, which of these values corresponds to a high frequency?

Adriaan
  • 17,741
  • 7
  • 42
  • 75
t.hicham
  • 45
  • 8

1 Answers1

2

fft2 gives a matrix consisting of complex numbers. These complex numbers have nothing to do with high or low freq components. To visualize the fft, you can apply fftshift to fourier transform matrix and if you image the absolute value of this shifted 2d fft graph, you can obtain the 2d frequency amplitute map similar to below:

enter image description here

Here, the center region belongs to low frequency components and high frequency components are outer regions from the center.

If you want to apply some filtering on high frequency components, you can create a circular binary image having the same center as the shifted fft image and set the outer regions to zero and apply ifft2 to get low pass filtered image

fatihk
  • 7,789
  • 1
  • 26
  • 48
  • Hi Mrs Angelina jolly. what that reflect the white lines in the shiftfft ? – t.hicham Apr 21 '16 at 13:38
  • @t.hicham they indicate that there are some frequent magnitute fluctuations in a particular direction with various periods, like zebra texture – fatihk Apr 22 '16 at 05:45
  • @t.hicham, for detailed explanation: http://homepages.inf.ed.ac.uk/rbf/HIPR2/fourier.htm – fatihk Apr 22 '16 at 05:47