Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!math.ohio-state.edu!cs.utexas.edu!news.sprintlink.net!tank.news.pipex.net!pipex!plug.news.pipex.net!pipex!dish.news.pipex.net!pipex!bbc!ras
From: ras@rd.bbc.co.uk (Richard Salmon)
Subject: Re: Problem with 3chip color CCD cam
Message-ID: <DD529u.942@bbc.co.uk>
Sender: usenet@bbc.co.uk
Nntp-Posting-Host: rdmailgate.rd.eng.bbc.co.uk
Organization: British Broadcasting Corporation, UK
X-Newsreader: TIN [version 1.2 PL0]
References: <40bf71$6mh@odo.PEAK.ORG>
Date: Fri, 11 Aug 1995 08:58:41 GMT
Lines: 42

Andreas Queisser (queisser@PEAK.ORG) wrote:
: I'm currently using a Sony XC003 color camera which has a three-way
: beamsplitter and 3 CCD sensors with color filters. When I look at
: images of white paper illuminated by "white" fluorescent light I can
: see that the top of the image is slightly red, the bottom slightly
: green. Looking at the individual channels I determined that the
: green channel is slightly lower than blue and red is slightly higher
: in the top region of the image. At the bottom the situation is reverse.
: Of course, when I do a white-balance at the top of the image I will get
: more green at the bottom with good white at the top.

: I tried different lenses and also tried not using a lens at all. I also
: compared the Sony to a Hitachi 3chip camera. The Hitachi showed similar
: results (maybe just a repackaged Sony.)

: At this point I'm wondering if 3 chip color cameras are actually well
: suited for doing accurate color measurement (I don't want to get into
: any colorspace discussions, the images are never displayed.)

Firstly, you'll not get any sensible colour measurements off a camera if
you're using normal flourescent lighting, and secondly, a camera is not
the best way to make colour measurement, since you're entirely at the
mercy of the camera's taking characteristics, matrix (if it's got one),
and gamma curve (NB camera gamma curves are not mathematical curves, but
have a variety of different shapes, with slope limited close to black).

The problem you seem to be describing sounds like a registation problem. 
If you've no control over registration on your camera, then I'm not sure
what to suggest.

For colour measurements, I've seen advertised the "Colortron".  Info is
available if you fax them in the US on (415) 461-8011.  It does proper
colorimetric measurements at 10nm intervals, and costs 1195 US Dollars,
including MAC software to drive it.  Has anyone had any experience of
this?  This sounds very good value compared to traditional
spectro-photometers.

--
Richard Salmon
--------------------------------------------------------------------------
richard.salmon@rd.bbc.co.uk       |   These are my personal views, and do
BBC R&D Dept, Kingswood Warren    |   not necessarily represent the BBC's.
