[GRASSLIST:3346] RE: r.le.pixel

[William L. Baker]

I thought Eric's idea would work, but on second thought
it may not.  When r.le.pixel encounters a null
value in the center of the window, it puts a null
value in the resulting moving window map where
the center pixel is.  If one puts some non-null
value around the edge of the map, that value will
enter into the calculation.  So, r.le.pixel in its
present form probably cannot be tricked into doing
the calculation.  Really, though, it is not
particularly meaningful to do the calculation when
the window is incomplete.  I can see it both ways.
Perhaps in the next version, I can add the option.

Bill Baker

-----Original Message-----
From: Eric G. Miller [mailto:egm2 at jps.net]
Sent: Monday, March 18, 2002 7:21 PM
To: grass list
Subject: [GRASSLIST:3345] RE: r.le.pixel


On Mon, Mar 18, 2002 at 07:05:04PM -0700, William L. Baker wrote:
> Hi David,
> 
> I'm the author of r.le.pixel.
> 
> r.le.pixel will leave a null row and
> column when you use a 3x3 window because
> the window starts centered over row 2 and
> column 2.  No value is defined for row 1 and
> column 1, since that pixel would not have the
> full 9 pixels.  The larger the window, the
> larger the null area along the edge. If you
> feel that a calculation should be done even
> though the windows would not be complete,
> you could add rows or columns as needed outside
> the image.  Not sure how to do this in GRASS,
> but it could probably be done.

Just increase the region.  Usually that will work, unless r.le.pixel
uses a raster's window rather than the current window (in which case,
you'll have to generate a new raster with the larger window setting).

-- 
Eric G. Miller <egm2 at jps.net>