On Mon, 2008-05-26 at 14:56 +0200, David Necas (Yeti) wrote:
> On Fri, May 23, 2008 at 12:15:23PM +0200, Renzo Stheins wrote:
> > Firstly, the statistics for the grains is incorrect, only 2 grains are
> > identified when using the inverted mask. 
> 
> I am not sure what `when using the inverted mask' means.  The mask
> defines the marked grains, in the sense that grains just *are*
> continuous parts of the mask.  (Grain marking and evaluation are
> completely separate operations since either can be done in many ways and
> one may also want to modify marked areas somehow between.)
> 
> If you invert the mask it is likely that it has only 2 continuous parts
> now, but then the statistics works as expected (for some value of
> expected at least).
> 
> > When disabling the invert of the mask, the number of grains is too high
> > (301), while a manual count yields 225.
> 
> While you might hit an unknown bug, I suppose this is more likely caused
> by different definitions of individual grain.  As I wrote above, the
> number of grains is the number of continuous parts of the mask that
> marks them (where pixels touching only with a corner are considered
> separate).
> 
> You can use the grain measurement tool
> 
>       
> http://gwyddion.net/documentation/user-guide/grain-analysis.html#grain-measure
> 
> to display the ordinal number of individual grains and thus check what
> parts constitute which grain.
> 
> > Is there some setting I can adjust to get the correct grain count and
> > size distribution?
> 
> Well, you need to get them marked according to your notion of a grain.
> The counting should work then.  It can be possible to achieve this with
> some adjustment of the watershed algorithm parameters.  If the algorithm
> produces tiny (noise) grains, these can be removed with the grain
> remove-by-threshold function.
> 
> > Secondly, when wanting to plot the distribution of the grains (with only
> > "Equivalent Disc Radius" selected) this results in a program crash. 
> > 
> > Selecting only other options, e.g. "Equivalent square side" also result
> > in a crash.
> > 
> > Exporting the raw data does work however.
> 
> I am afraid this is a long known bug.  It was found and fixed almost
> immediately after the release of 2.9, the patch is here:
> 
>       
> http://gwyddion.svn.sourceforge.net/viewvc/gwyddion/trunk/gwyddion/modules/process/grain_dist.c?r1=8455&r2=8664&view=patch
> 
> However, this can help you only if you build Gwyddion from source code.
> The next version will have this bug fixed.

If you don't want (can't) recompile Gwyddion for Windows you can also
downgrade GTK+ to, I think, version 2.8 (I can't check right now). You
have to go down 1 or two numbers from the latest release. This worked
for our windows installations. In Linux of course recompiling is
probably easier.

Gregor

> 
> Yeti
> 
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________
> Gwyddion-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/gwyddion-users

-- 
!!I am trying to shift my personal mail to this address!!
                plz update your addressbook

 Gregor Hlawacek  --  http://www.unileoben.ac.at/~hlawacek
 Proleber Straße 60/7        --  +43 650 3592145
 8700 Leoben, Austria
                (o_     (o_     (o<)<   (o< -!
                //\     //\     //\     //\
                V_/_   [ ]/_    V_/_    V_/_
                penguin         eating fish
                        travelling      being noisy
        GPG ID: FBCDF36E
 GPG Finger Print: 1AA0 DC5C 186F 6144 01F0  8FB2 7ABD D7FB FBCD F36E


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Gwyddion-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gwyddion-users

Reply via email to