Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!scramble.lm.com!news.math.psu.edu!news.cac.psu.edu!news.cse.psu.edu!uwm.edu!newsfeed.internetmci.com!in1.uu.net!bcstec!nntp
From: Mike Fouche <mikef@hsvaic.hv.boeing.com>
Subject: Re: Image Classification Based on Data Reduction.
X-Nntp-Posting-Host: e570254.hv.boeing.com
Content-Type: text/plain; charset=us-ascii
Message-ID: <DqzBBC.623@bcstec.ca.boeing.com>
Sender: nntp@bcstec.ca.boeing.com (NNTP News Access)
Content-Transfer-Encoding: 7bit
Organization: Missiles & Space Division
References: <4m6t7c$lvc@ssv2.dina.kvl.dk>
Mime-Version: 1.0
Date: Mon, 6 May 1996 10:18:47 GMT
X-Mailer: Mozilla 1.2 (Windows; U; 16bit)
Lines: 18

Hello,

If you're trying to classify these images rapidly in real time then it
can be a problem - although there are ways to get around this problem.  
Otherwise perhaps the processing time is not so important.

We haven't used PCA but we have used SVD, wavelets, and FFTs as filters 
or pre-processors for image processing.  The wavelets are the fastest 
in terms of processing time.  Typically a 65,536 pixel image is reduced
to 32 or so "representative" numbers for input into a neural net.  We've
found this to work very well across several image processing 
applications.  You might try one of these.  One last comment - SVD works
so the best (so far) in terms of it's ability to extract the target 
features.

Mike Fouche
Boeing Missiles & Space Division

