Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!udel!gatech!swrinde!howland.reston.ans.net!news.sprintlink.net!EU.net!sun4nl!news.nic.surfnet.nl!tudelft.nl!deneb!klamer
From: klamer@ph.tn.tudelft.nl (Klamer Schutte)
Subject: Re: Looking for accurate 2D interpolation algorithms.
Message-ID: <klamer.791390691@deneb>
Lines:       32
Sender: news@news.tudelft.nl (UseNet News System)
Nntp-Posting-Host: deneb.ph.tn.tudelft.nl
Organization: Delft University of Technology
References: <3g4o9p$d8g@atom.ansto.gov.au> <AZ.95Jan27161712@saturn.analog.com>
Date: Sun, 29 Jan 1995 14:44:51 GMT

In <AZ.95Jan27161712@saturn.analog.com> az@saturn.analog.com (Alex Zatsman) writes:

:In article <3g4o9p$d8g@atom.ansto.gov.au> rfulton@atom.ansto.gov.au (Roger Fulton) writes:


:>    I'm looking for accurate 2D interpolation algorithms.
:>    A bicubic algorithm I'm using (which is built in to IDL) is good,
:>    but not quite good enough for my needs. It produces values
:>    which are greater than any that were present before
:>    interpolation.

:Are you sure this is a problem? For example, if in 1-dim case you have
:values  [1,2,3,3,2,1] at points  [0,1,2,3,4,5] then at  point 2.5 most
:interpolation schemes would produce a value larger than 3.

:>    Does anyone know of any other algorithm which may perform better
:>	than bicubic ? 

:It really depends on what "better" means. If the above condition is
:your criterion, then in 1D case you would use linear interpolation,
:but I'm not sure what is a good analogy in a 2D case.

Bilinear interpolation. This is a linear interpolation in the first
direction followed by a linear interpolation in the second direction.

I believe it is described in Numerical Recipes in C, second edition. 

Klamer
-- 
Klamer Schutte	-- +31-15-786054 / +31-15-692000
klamer@ph.tn.tudelft.nl / schutte@tpd.tno.nl
http://www.ph.tn.tudelft.nl:2000/People/klamer/Klamer.html
