Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!hao!ames!hc!beta!jpm
From: jpm@beta.UUCP (James P Mcgee)
Newsgroups: comp.graphics
Subject: Re: Data Compression
Message-ID: <12458@beta.UUCP>
Date: Tue, 17-Nov-87 13:20:20 EST
Article-I.D.: beta.12458
Posted: Tue Nov 17 13:20:20 1987
Date-Received: Thu, 19-Nov-87 21:59:39 EST
References: <619@applix.UUCP> <305@etn-rad.UUCP> <3387@adobe.COM>
Organization: Los Alamos Natl Lab, Los Alamos, N.M.
Lines: 21
Keywords: compression fonts
Summary: Glenns idea = Tom Stockhams vector quantization


   There were a couple of messages about compression this past week that
were related, but the relation was not obvious.  Spencer Thomas mentions
the vector quantization methods of Tom Stockham.  Later, Glenn Reid
of Adobe describes an idea for a compression mechanism.  The idea that
Glenn came up with is very similar to Stockhams vector quantization.  The
difference is that, while Glenn suggests using standard tables of bitmaps,
Stockham uses tables of bitmaps generated from the picture.  Including
the bitmaps doesn't increase the size of the compressed image appreciably,
and the results are better.  My complements to Glenn.  Even though it is
a little too little and a little to late to be first, it's definitely a
creative idea.

   As an aside, I met Tom Stockham when he gave a talk for Hank Christiansens
MOVIE.BYU training program.  In his talk, Tom derived the theoretical 
minimum number of bits necessary to encode any image, which is 70 bits.
This is 70 bits for the entire image, regardless of size, resolution, or
content.  The derivation is interesting, humerous, and gives absolutely no
guidance on how to implement such a scheme.

Pat McGee, jpm@lanl.gov