Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!iuvax!ndcheg!uceng!mfinegan
From: mfinegan@uceng.UC.EDU (michael k finegan)
Newsgroups: comp.graphics
Subject: color quantization / equalization ==> 256 colors
Summary: Are there any PORTABLE routines about ?
Keywords: color, quantization, 256, image processing
Message-ID: <1148@uceng.UC.EDU>
Date: 6 Jun 89 01:08:07 GMT
Followup-To: mfinegan@uceng.uc.edu
Organization: College of Engg., Univ. of Cincinnati
Lines: 24
References:


    I have tried to compile, and run, several of the color quantization programs
which are available via FTP, and have been discussed in comp.graphics. Some
(fbquant from fbm ...) require the toolkit to be compiled, and won't compile
(easily :-< ) under SysV (AT&T 3B2); others (Steinberg annotated routine) are
missing header files (a problem since that's where all the structures are
defined :->); still others are extrordinarily inefficient, and use euclidean
distance to map the colors ... Does anyone have references for all required
articles (1 for median cut, 1 for dithering, etc.) to allow the routine to be
written easily (i.e. algorithmic discussion)?

    I would like to run the routine on a pc, so portability is a consideration.
As the display of color images on the VGA display requires a different LUT
for each image, efficiency is a consideration.

    Would mapping to a different color space, and then quantizing each component
(i.e. truncate least signifigant bits), or then equalizing each component
separately (JPL?), be faster/better than 3-D approaches?

    Could you share your opinions, or experiences ?

				Thanks,
					Mike Finegan
					mfinegan@uceng.uc.edu