Analyzing a Fourier-Transformed image, getting ridge effects despite windowing

Hello all,

I'm still trying to analyze a thermographic picture of a carbon fibre reinforced plastic object.

Right now I'm attempting to gauge the direction of lines in the image by taking the 2D Fourier Transform (fft2) of the picture, then going through the image in a half-circle around the center and summing up the amplitude values of all pixels.

I've already gotten it to work for a picture of fibres that I got from a different research paper, and, as you can see below, I get a very clean Fourier image with four maxima, corresponding to the two directions of fibres from the test image, just by applying a basic Hann window.

Now, attempting to do the same with my own images, I run into problems with what I think are edge effects.

Steps I took to achieve this picture:

1) take image, cast to double, remove mean

2) multiply with window function from this window generator tool

3) take fft2 of the windowed image, multiply it with its complex conjugate to remove the imaginary part, and then plot it.

Those are the same steps that produced a good image with the first image set.

Does anyone have any ideas for why

1) I get those trailing "comet trails" in my own images?

2) the maxima in my own images are way closer together than the maxima in the picture above, even through my lines are far closer together?

Thanks in advance!