[Gwyddion-users] pattern recognition

[s***@bo.ismn.cnr.it]

Good morning everybody,

my question is the following;

I have a high resolution AFM image where a single feature, let's say a
triangle-like shape, is repeating periodically. No matter in which
directions.

Of course, since we are talking about raw AFM data, not all the triangles
are perfect, some are properly visible, some are a bit shady and so on.

What I would like to do with Gwyddion is to select the best triangle (say
with a crop) and repeat THAT specific best triangle in the positions of
all the other less good triangles.

I guess the procedure should be the following:

1) select with a crop the best triangle

2) perform a 2d-correlation to find the positions where I have all the
triangles

3) "substitute" to each peak position in the 2d-correlation, the best
triangle.

Is all this procedure achievable in Gwyddion?

Thank you a lot for the help!!!

Stefano
--
Dr. Stefano Chiodini, PhD
Consiglio Nazionale delle Ricerche (CNR)
Istituto per lo Studio dei Materiali Nanostrutturati (ISMN)
Via Piero Gobetti 101
I-40129 Bologna
I T A L Y

Skype: stefano.chiodini6
***@bo.ismn.cnr.it
Phone:++39-051-639-9188
Cell: ++39-3393142341

www.bo.ismn.cnr.it


------------------------------------------------------------------------------

[Daniil Bratashov]

On Fri, 16 Oct 2015 09:09:48 +0200

Post by s***@bo.ismn.cnr.it
Good morning everybody,
my question is the following;
I have a high resolution AFM image where a single feature, let's say a
triangle-like shape, is repeating periodically. No matter in which
directions.
Of course, since we are talking about raw AFM data, not all the
triangles are perfect, some are properly visible, some are a bit
shady and so on.
What I would like to do with Gwyddion is to select the best triangle
(say with a crop) and repeat THAT specific best triangle in the
positions of all the other less good triangles.
1) select with a crop the best triangle
2) perform a 2d-correlation to find the positions where I have all the
triangles
3) "substitute" to each peak position in the 2d-correlation, the best
triangle.
Is all this procedure achievable in Gwyddion?
Thank you a lot for the help!!!
Stefano

Similar procedure is implemented in Data process->Correct
data->Correlation averaging, but this module is a kind of experimental
one, so subpixel position of correlation maximum, that was made
available recently in API, probably is not used there.

WBR, Daniil Bratashov.

------------------------------------------------------------------------------