Path: news.mathworks.com!not-for-mail
From: <HIDDEN>
Newsgroups: comp.soft-sys.matlab
Subject: Re: Embedded Image Resizing using FFT
Date: Thu, 20 Nov 2008 20:26:02 +0000 (UTC)
Organization: Oxford University
Lines: 7
Message-ID: <gg4h4q$7oc$1@fred.mathworks.com>
References: <gg4co5$2cu$1@fred.mathworks.com>
Reply-To: <HIDDEN>
NNTP-Posting-Host: webapp-05-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1227212762 7948 172.30.248.35 (20 Nov 2008 20:26:02 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Thu, 20 Nov 2008 20:26:02 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1073021
Xref: news.mathworks.com comp.soft-sys.matlab:502124


"Kevin Voet" wrote:
> I'm currently looking for a way to rescale/resize images using Embedded Matlab. 
> I've read it should be possible using the Frequency domain. And since the fft2 and ifft2 functions are supported by the Embedded Matlab subset I thought I'd go this way. 

This has often struck me as an excellent way to resize images. You crop the image in the frequency domain, getting rid of the high frequencies that cannot be seen in the smaller image, then transform back to the spatial domain. It should produce the 'ideal' resized image, i.e. one that retains the maximum amount of information possible in the smaller size.

I've tried to do this myself, cropping off various different parts of the frequency image, but never got back anything that looked right. The question is, can you just cut out bits of the frequency image, then transform back, or is there some reason why this cannot work? If it is possible, which parts do you cut out?