From: Tom Clune <t_clune@medispectra.com>
Subject: Re: image processing with RGB
Date: 30 Apr 1999 00:00:00 GMT
Message-ID: <7gc93d$khu$1@nnrp1.dejanews.com>
References: <37288EF9.F8BEBEBF@polonium.de> <7gad22$1qb$1@nnrp1.dejanews.com> <3728D656.2262343C@polonium.de>
X-Http-Proxy: 1.0 SBS1, 1.0 x2.dejanews.com:80 (Squid/1.1.22) for client 209.6.173.190
Organization: Deja News - The Leader in Internet Discussion
X-Article-Creation-Date: Fri Apr 30 12:53:02 1999 GMT
Newsgroups: comp.graphics.algorithms
X-Http-User-Agent: Mozilla/4.0 (compatible; MSIE 4.01; Windows NT)

In article <3728D656.2262343C@polonium.de>,
  Rafal <rafcio@polonium.de> wrote:
> Tom Clune wrote:
>
> > In article <37288EF9.F8BEBEBF@polonium.de>,
> >   Rafal <rafcio@polonium.de> wrote:
> > > Hallo,
> > >
> > > I wont to use my filters on colored digital images. But all algorithms
> > > which I have, are for grayed images. How can I use this algoriths for
> > > colored images?
> > >
> >
> > I don't know how other people do it, but I always convert to YUV color
space,
> > and only filter the Y (grayscale) data. After filtering, you can switch back
> > to RGB if you like. Trying to filter color images directly can give you some
> > funny artifacts in the color.
>
> Thanks. In my book I found only YIQ color space. Can you send me the
> matrix for converting from RGB -> YUV. Or ist YUV equal to YIQ?
>

I'm not familiar with YIQ. But if the Y is just luminance, it should do.
Here's my comment from the code I use to do the conversion. R is the red
color value, G is the green color value, and B is the blue color value. All
colors are 8- bit. If your color depth varies, change the 128 value
accordingly. (BTW, YCbCr == YUV).

/* calculate the YCbCr pixel data and copy it to m_YData, m_CbData, and
m_CrData. The factors
   that I use to convert to luminance and chrominance are from CCIR
Rec. 709:
   Y = 0.2125R + 0.7154G + 0.0721B;
   Cr = (R - Y) + 128
   Cb = (B - Y) + 128
*/

FWIW.

--Tom Clune, MediSpectra, Inc.

-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/       Search, Read, Discuss, or Start Your Own