Box Counting Method and r.le

[William L. Baker]

Hello,

Regarding r.le...First, I have not been able to keep
up with revisions to r.le recently, and I apologize for
this, but do not have an immediate solution.  Second,
r.le does not have a box-counting algorithm for calculating
fractal dimension of images.  There are reasons to worry
that fractal dimension is not a good measure to use with
images (see Frohn, Robert C. 1998.  Remote sensing for
landscape ecology: new metric indicators for monitoring,
modeling, and assessment of ecosystems. Lewis Publishers,
Boca Raton, Florida).  However, a program that is aimed
more specifically at remote sensing imagery, and I think
does include fractal dimension using box counts is called
ICAMS, and is by Dale Quatrocchi (spelling?).  Not sure
where to get it, but it's out there somewhere.

In general, if you are having problems with r.le there are
instructions in the documentation about how to report them.
Most useful to me is to know details of your platform, the
maps used, and especially whether you observe any warnings
or errors during compilation.  I will try to help as I can,
but have limited time at the moment.

Bill Baker
University of Wyoming
BAKERWL at UWYO.EDU
Original author of r.le

-----Original Message-----
From: DIETER LEHMANN [mailto:lehmann at fh-nuertingen.de]
Sent: Thursday, September 23, 1999 7:12 AM
To: grass at cecer.army.mil
Subject: Re: Box Counting Method


Hi Markus,
I think, that r.le should do the job. But - it is not working for me (I 
send a mail to the list some time ago, but received no answer...)

Dieter



From:           	neteler at geog.uni-hannover.de (Markus Neteler)
Subject:        	Re: Box Counting Method
To:             	grass at cecer.army.mil
Date sent:      	Fri, 17 Sep 99 7:32:36 BST
Send reply to:  	grass at cecer.army.mil

> > I've written a box counting program for GRASS.  It uses bit-flipping
> > techniques which means that it's very quick, but as a result it may
> > only run on little endian machines (e.g. Intel-based machines, which
> > is all I have access to).  You can find out more and download the
> > source code by following the links from my signature.
> [...]
> > Dr. Mark Lake
> > http://www.ucl.ac.uk/archaeology/research/profiles/lake.htm
> > 
> 
> Hi Mark, hi all,
> 
> currently I am searching for a GRASS module to calculate
> the fractal dimension of remote sensing images. I think
> the box counting method should be fine for that.
> So I have a few questions (as not being very experienced with fractals
> in GIS yet):
> 
>  - I would need a module which accepts grey scale images instead
>    of binary images
>  - the result should be regionalised: A resulting raster map would show
>    the distributed fractal dimension so that you could distinguish
>    different types of landscape
> 
> Is that possible with (modified) r.boxcount? Or do I have to write a
> new module? 
> Does anybody have experience here?
> 
> Any help is appreciated!
> 
>   Markus Neteler


_____________________________________

Dieter Lehmann 
State University of applied Sciences
Schelmenwasen 4-8
D-72622 Nuertingen, Germany
lehmann at fh-nuertingen.de
Phone: +49 (0) 7022 404 Ext. 192 or Ext. 152
Fax: +49 (0) 7022 404 Ext. 166