Re: Is there any way to solve the aliasing of an image in the spatial domain

As you sub-sample the DFT spectrum envelope, you loose some fine
details of it. When you translate the new spectrum back to the new
image, this translates to aliasing effects similar to the ones that
appear in mpeg and jpeg files with very low quality (high compression
level).

In order to avoid aliasing, you either make sure your sub-sampling does
not destroy the profile of the original spectrum (i.e. make sure you
use enough samples, usually more than 1/4 of the original size), or
decide to sub-sample the image itself and keep only the most
significant features (i.e. cut the higer frequencies and keep all the
lowers one that usually contain most of the image content).


--
Harris

.