Strange FFT spectrum of background noise

Posted by jumpfunky on
URL: http://imagej.273.s1.nabble.com/Strange-FFT-spectrum-of-background-noise-tp5011253.html

Dear list,

I currently have some TEM images with a strange background noise pattern
(background.png). It seems to be "red shifted" noise.  White noise
exhibit the same power for each frequency. Red shifted noise exhibed
lower power for higher frequencies. You can reproduce this kind of
background noise if you create a x*y image, then add gaussian noise and
upscale (bilinear interpolation) it to a (3*x)*(3*y) image. So my guess
is, that the TEM camera simply upscale lower resolution to a higher
resolution image.

But there is one discrepancy:
The fft spectra of these simulated image (spectrumSimulated.png) is
different from the fft spectrum of the background of the TEM image
(spectrum.png). The TEM spectrum shows a radial symmetric sinc pattern
whereas the simulated spectra shows a rough sinc pattern in x and y
direction.

Any idea which interpolation technique can produce this radial symmetric
sinc pattern in the fft spectrum?

Another idea: The sinc function is an ideal reconstruction filter. It is
possible that the camera simply apply a reconstruction filter which
leads to this kind of fft spectum?

Cheers,
Thorsten

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html