Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!samsung!munnari.oz.au!comp.vuw.ac.nz!canterbury!cosc.canterbury.ac.nz!raymond
From: raymond@cosc.canterbury.ac.nz (cantva)
Newsgroups: comp.dcom.fax
Subject: Re: Compressing a B/W Fax image
Message-ID: <1991May6.100642.621@csc.canterbury.ac.nz>
Date: 5 May 91 22:06:40 GMT
References: <1991May5.160009.16327@cs.dal.ca>
Reply-To: raymond@cosc.canterbury.ac.nz
Lines: 36
Nntp-Posting-Host: ruru.cosc.canterbury.ac.nz

From article <1991May5.160009.16327@cs.dal.ca>, by gauthier@ug.cs.dal.ca (Paul Gauthier):
> I'm interested in what the optimal method for compressing B/W fax images

Are you interested in compression, speed, or both?

> I'm interested in pointers
> to code that I can use or documents which would detail the method.

Depends on above. Adaptive arithmetic for compression, run length for speed.

> Also, what kind of compression ratio/effectivness/etc should I expect
> from the methods you suggest? (On a single page image 1728x2200 [Fine] or
> 1728x1100 [Standard]).

Run length will yield about 7:1 compression (see G3 standard for more accurate
assessment).  A good adaptive arithmetic compressor will rarely do worse than
20:1.

I have built an adative arithmetic compressor to compress fax images. It is
resolution independant and will handle arbitrarily large fax images (as long
as the memory holds out!).

The test images I tried were about 1700x2300 (over an A4 page) (513216 bytes).
The worst it has done is 20.5:1 (~25000 bytes) (full page of (dense) text) 
and the best is 171:1 (< 3000 bytes) (simple digram). 

These two are in the CCITT fax test documents (I forget the numbers).


Raymond.

--
Raymond Wilson.	email:	raymond@cosc.canterbury.ac.nz
		snail:	c/- Computer Science Department,
			University of Canterbury,
			New Zealand.