Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!lll-winken!uunet!tektronix!nosun!fpssun.fps.com!celerity!hutch
From: hutch@celerity.uucp (Jim Hutchison)
Newsgroups: comp.graphics
Subject: Re: Subroutine for NTSC to YIQ/RGB ?
Message-ID: <199@celit.UUCP>
Date: 25 Apr 89 15:51:01 GMT
References: <883@uceng.UC.EDU>
Sender: news@celerity.UUCP
Reply-To: hutch@celerity.UUCP (Jim Hutchison)
Distribution: na
Organization: FPS Computing
Lines: 26

In article <883@uceng.UC.EDU>:
>[...] convert digitized NTSC signal to either YIQ or RGB signal.
>The NTSC was sampled at 4 times the sub-carrier frequency and has values 
>from 0 to 255; alternate lines are 180 degree out of phase. Any information
>or reference to book, paper concerning this would also be appreciated.

Looking at NTSC, you get the luminance (Y) from a low frequency component of
the signal and chroma (I and Q) from the frequency components.

                                  <- white
                 /| /\
            /\/\/ |/  \/\/\
-+  +-||||-'               `-+    <- black
 +--+                        +-
	   |---- image ---|

               --time-->

As to your getting the I and Q values back out of the samples, maybe that
"new" cosine encoding the "1.2 bits per video pixel" folks are up to would
get you values of theta which you could use.  The function I am refering
to is like a fourier transform except that it works with angular data such
as I and Q.  Any comments folks?

/*    Jim Hutchison   		{dcdwest,ucbvax}!ucsd!celerity!hutch  */
/*    Disclaimor:  I am not an official spokesman for FPS computing   */