Just a question: why do you normalize the MTF on the second value in the array?

[sevbogae]

In the normalization of the MTF, you normalize on the second element:

tmp = ft_fit[1];
for (i=0; i<ft_len; i++) {
  ft_fit[i]= ft_fit[i]/tmp;
}

Why not ft_fit[0]?

I suppose your comment has something to do with it, but I don't really see what you mean...
// imagej FT is squared sum of positive negative spatial frequencies, except for FT(0)

[emx77]

For some reason ft_fit[0] has a too low value which is probably be related to normalization and the DC/mean component in the image,
The imagej FFT is briefly described here:
https://wsr.imagej.net/macros/examples/TestArrayFourier.txt
As I understand it, The FFT is a function of the RMS of positive and negative frequencies. I hope this helps you further.