FFT spectral overlap

While working with the FFT I noticed that some energy appears that does not radiate from the FFT center.

Is this energy at frequencies above nyquist?

Original Image___________________FFT________________________ FFT copy

Does anyone have any thoughts on how to handle this?

rrd

Good day Ron!

Is this energy at frequencies above nyquist?

Yes!

For discrete Fourier transformations the planes must be imagined on a torous or periodically repeated.

Your image data isn’t correctly sampled (Shannon theorem).

This happens if the objects you’ve imaged contain high frequencies and your imaging device doesn’t provide the necessary lowpass.
It may also happen in case of synthetic images that are most often not correctly bandlimited. Think of an image consisting of a single pixel different from zero. Such an image has an infinitely extending Fourier-spectrum.

What you see “spectrally folded in” are so-called alias-components that generally can’t be removed post hoc.

HTH

Herbie
PS:
I must admit that I can’t reproduce the Fourier spectrum of the provided image (looks like a photograph of a rotating grating)!
What I get as the log Power Spectrum is:

1 Like

Hi Herbie

How did you get that great FFT?

The original image is a blending of two sinusoidal synthetic waves at 15 degrees to one another…the exact angle you get in your fft.

image

I then use the built in FIJI FFT.after DCfree windowing (your plugin)

What are the steps you used to get your log power spectrum?

Thanks
ron