Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!apple!bionet!agate!riacs!hitchner
From: hitchner@riacs.edu (Lewis Hitchner)
Newsgroups: comp.graphics
Subject: Re: Fast Image Scaling
Message-ID: <1990Nov6.123207@riacs.edu>
Date: 6 Nov 90 20:32:07 GMT
References: <A1346770260@syntel.syntel.mn.org> <1990Oct26.011905.16187@morrow.stanford.edu>
Sender: hitchner@riacs.edu (Lew Hitchner)
Reply-To: hitchner@riacs.edu (Lewis Hitchner)
Organization: Research Institute for Advanced Computer Science
Lines: 13


That should hardly be called "image scaling".  All you are doing is sub-sampling,
not scaling.  Thus, you are also incurring all the bad aliasing effects caused
by reducing the sample rate of a non-band limited image.  Try your algorithm
on a checkerboard image and "scale" your result by 1/2n where n is the no. of
pixels in each block of the checkerboard.  You might want to refer to any of the
standard textbooks on digital signal processing or on digital image processing
for a complete explanation of correct scaling methods (most of these texts are
named "Digital Signal Processing" or "Digital Image Processing" with minor variations on those names -- popular textbooks are by the authors: Pratt,
Gonzales & Wintz, Anil Jain, Castleman, Oppenheim & Schafer, Rosenfeld & Kak).

	Lew Hitchner
	RIACS at NASA Ames