Interpretation of results using directionality plugin

[anusuya pal]

Hello,

I have a trouble in interpreting the results which I have using
directionality plugin. For the convenience, I am attaching the image, in
which I want to find all possible information related to the cracks.
Firstly I want to find the orientation of the cracks, and that's why I have
used directionality plugin to the gray-scaled image. Is this the correct
way to use this plugin, or I need to change the image so that it only has
cracks (The variance filter is used on the gray-scaled image, and attached
as modified.jpg)? I went through the details given on
http://imagej.net/Directionality

The results using local gradient orientation on the gray-scaled image are-
direction -90.77
dispersion  1.03
amount 0.07
goodness 0.95

As my advisor doesn't have much idea about ImageJ, it's very hard for me to
do the interpretation of the above results. Whatever I understood is the
following.

As the result gives the flat histogram mostly, there is no preferred
direction of the structures except the one at around -90 and 90 degrees
though I don't understand what does that peak means.

The Center of the Gaussian is given by -90.77 with the standard deviation
of 1.03. (The fit is good as the goodness is close to 1). The proportion of
the structure orienting at around -90 is 7%. As the amount value is 0.07,
it implies about the uniformity of the structures.

May I request to go through the results?

Thanks
Anu

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Herbie]

Dear Anu,

attached please find two results that I obtained with code of mine.
You can find a description of the approach here:
<www.gluender.de/Writings/WritingsTexts/HardText.html#Gl-2013-1>

--Orientation Salience of "aaa".png--
This result is obtained from the windowed excerpt
"aaa_d1949s390x6y76.png" of your original image.
The analysis excludes spatial frequencies lower than 2% of the Nyquist
frequency.

--Orientation Salience of "modified".png--
This result is obtained from the windowed excerpt
"modified_d1949s390x6y76.png" of your preprocessed image.
The analysis uses the full spatial frequency range.
I'm not sure whether your preprocessing is adequate.

Both result plots are comparable, i.e. they show a dominant orientation
near 180deg and another strong orientation around 145deg.
(In the plots the angle starts horizontally and increases
counter-clockwise.)

Concerning the plugin "Directionality" when used in the Fourier-mode, I
must confess that I can't follow the author's approach and I judge it
not being state of the art.

Finally, I should like to mention that global/l / l/regional orientation
salience analysis, as explained in the cited report, is far from trivial.

Regards

Herbie

::::::::::::::::::::::::::::::::::::::::
Am 15.10.17 um 22:52 schrieb anusuya pal:

> Hello,
>
> I have a trouble in interpreting the results which I have using
> directionality plugin. For the convenience, I am attaching the image, in
> which I want to find all possible information related to the cracks.
> Firstly I want to find the orientation of the cracks, and that's why I have
> used directionality plugin to the gray-scaled image. Is this the correct
> way to use this plugin, or I need to change the image so that it only has
> cracks (The variance filter is used on the gray-scaled image, and attached
> as modified.jpg)? I went through the details given on
> http://imagej.net/Directionality
>
> The results using local gradient orientation on the gray-scaled image are-
> direction -90.77
> dispersion  1.03
> amount 0.07
> goodness 0.95
>
> As my advisor doesn't have much idea about ImageJ, it's very hard for me to
> do the interpretation of the above results. Whatever I understood is the
> following.
>
> As the result gives the flat histogram mostly, there is no preferred
> direction of the structures except the one at around -90 and 90 degrees
> though I don't understand what does that peak means.
>
> The Center of the Gaussian is given by -90.77 with the standard deviation
> of 1.03. (The fit is good as the goodness is close to 1). The proportion of
> the structure orienting at around -90 is 7%. As the amount value is 0.07,
> it implies about the uniformity of the structures.
>
> May I request to go through the results?
>
> Thanks
> Anu
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html