Wed Feb 08 10:31:46 CET 2012
You need MATLABŪ to run this scripts by your own.
Original image Duck.png
Same image but grayscaled.
Gaussian filter appended
Download script1.m
Invoke with:
Invoke with:
script1('Duck.png')
Unsharp masking operation
Download script2.m
Invoke with:
script2('Duck.png')
Sharp filtered
Download script3.m
Invoke with:
script3('Duck.png')
Periodical interferences occure, if some frequences are overlayed by other frequences. Those frequences can be good spotted by looking at the fourier transformed image. Dark regions of the fourier specturm indicates no or very low frequences, bright regions high frequences. The disturbing frequences can be removed by setting them to zero. This is achieved in the java application iFilter by painting over all bright white spots (except the centered main spot) in the fourier transformed image. An inverse fourier transformation of the modifed fourier spectrum lets you see the corrected image. The results depend on the users mouse skills. Try it by yourself!
Nasty interferences all over a pretty clown face.
The clown's fourrier spectrum shows the interferences (white stars).
My feeble attempt to paint over them.
Much better now.
Download the zip-archive that contains iFilter & 4 images to filter here.