Path: utzoo!attcan!uunet!snorkelwacker!usc!zaphod.mps.ohio-state.edu!ncar!unmvax!pprg.unm.edu!krukar
From: krukar@pprg.unm.edu (Richard Krukar [CHTM])
Newsgroups: comp.graphics
Subject: Re: Compression of multi-bit images
Message-ID: <27663@pprg.unm.edu>
Date: 11 Jun 90 19:07:32 GMT
References: <8099@b11.ingr.com> <SPENCER.90Jun4170821@spline.eecs.umich.edu> <1990Jun7.114905.1714@athena.mit.edu>
Reply-To: krukar@pprg.unm.edu (Richard Krukar [CHTM])
Followup-To: comp.graphics
Organization: U. of New Mexico, Albuquerque
Lines: 17


	I worked with various images compression schemes a few years ago
and I found a very disgusting method.  Two steps:
    1) Write up your favorite predictor.
	example: err(row,col) = images(row,col) - images(row, (col-1))
       and run it on your image to obtain an image of prediction errors.
       Don't pack the bits.  Store each number in bytes, shorts, words.

    2) Run compress ( LZW compression ) on the error image.

	This method is fast, lossless and very easy to implement.  It will
also give you a way to measure your own methods performance.  This technique
shamelessly trashed a number of multi-resolution based, tree based, fractal
based, DCT based, ....

	Regards
	Richard Krukar  (krukar@pprg.unm.edu)