Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!syntron!orcisi!michael
From: michael@orcisi.UUCP
Newsgroups: comp.graphics
Subject: Re: A few questions [really image quantization]
Message-ID: <782@orcisi.UUCP>
Date: Tue, 20-Jan-87 13:33:32 EST
Article-I.D.: orcisi.782
Posted: Tue Jan 20 13:33:32 1987
Date-Received: Tue, 20-Jan-87 23:49:44 EST
References: <563@vu-vlsi.UUCP> <476@pixar.UUCP>
Organization: Optical Recording Corporation, Toronto, Ontario
Lines: 12

This isn't a topic I claim to know very much about but it seems that an
exact histogram of the 24-bit image would be useful for determining the
"colour content" of an image.

One way of doing this is to treat the pixels in your frame buffer as a
linear list of integers and (quick)sort them.  After you run-length
encode the sorted image, the result represents an exact histogram of
the original image.

I had microcoded the quicksort on the Adage/Ikonas 3000 at the
University of Waterloo and it took a couple of seconds for the sort to
complete and was entertaining to watch.