Newsgroups: sci.image.processing
From: Steve@dstrip.demon.co.uk (Steve Rencontre)
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!gatech!howland.reston.ans.net!pipex!peernews.demon.co.uk!dstrip.demon.co.uk!Steve
Subject: Re: Image file formats and compression
References: <MXM.95Apr24110841@dcs.ed.ac.uk> <3niu6p$f4a@ed.petech.ac.za>
Organization: Datastrip Limited
Reply-To: Steve@dstrip.demon.co.uk
X-Newsreader: Newswin Alpha 0.7
Lines:  28
X-Posting-Host: dstrip.demon.co.uk
Date: Wed, 26 Apr 1995 14:45:17 +0000
Message-ID: <252339006wnr@dstrip.demon.co.uk>
Sender: usenet@demon.co.uk

In article: <3niu6p$f4a@ed.petech.ac.za>  9518132@ml.petech.ac.za (Tom Lessing) writes:

> and look for the Pcx document file the PCX format is a lossles compression 
> using an RLE mode for encryption (I find it to be usually the best of the all the formats
> but it depends on the image's used to test the format.

PCX is *very* unlikely to give good results on continuous-tone greyscale images.

All of the 'good' compression systems for these images are lossy, even though 
the error may be negligible, or less than the noise inherent in the image. About 
2:1 is the best you can expect from a standard lossless compressor like pkzip or 
gzip, and you could get near-as-dammit no compression, depending on your image.

There is continuing research into optimised lossless methods for images,
but a few lsb's worth of noise in a typical image can really screw up any
algorithm that has to reproduce the image with bit-for-bit accuracy.

IMO, the only *real* need for 100% bit-for-bit accuracy is as insurance
against being sued, as I hear is happening in the US over compressed medical
X-rays. 


-----------------------------------------------------------------------
Steve Rencontre               |  steve@dstrip.demon.co.uk (business) 
If it works, it's obsolete.   |  steveren@cix.compulink.co.uk (private)
-----------------------------------------------------------------------

