Best way to correct to circular particles

[Adam Hughes]

Hi,

I am using imagej to count and size nanoparticles, and the SEM images that
I produce often have non-circular boundaries due to both imperfections in
the images from the mictroscope, and from added noise when we do the
thresholding.  I've attached a few images that show some of these
imperfections.  Sometimes the particles are close to circular, sometimes
they have rough edges, and sometimes they are almost triangular.

I'd like to come up with a way to estimate the equivalent circular
particles from the outlines assigned to the black and white particles.
 What I'm doing now is just taking the total area and computing r as if the
total area was that of a circle.  Can anyone think of a more sophisticated
approach?

Thanks.

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

[John Dunsmuir]

Adam,
           Do any of the shape descriptors in the set measurements menu work for you?  I'm not sure what you mean by equivalent circular particles since you can obtain a radius from either the area or perimeter.  The documentation explains it better.

          Or are you trying to segment the image into a collection of roundish features?  Voronoi or watershed may help.  I'm currently using a macro I wrote to identify non-interpenetrating circles to identify closely spaced features that are assumed to be round.

John D.

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

[Adam Hughes]

Hey John,

It's kind of hard to know which shape descriptors would be ideal.  I guess
the best for me to do is to look at as many of the images as I can and see
which shape descriptors may correspond best to the deviation I'm seeing.
 My deviations, for example, may be more suited to one correction over
another.  I don't think I thought the question through before posting, and
for that I apologize.

Thanks


On Wed, Apr 17, 2013 at 6:35 PM, John Dunsmuir <[hidden email]> wrote:

> Adam,
>            Do any of the shape descriptors in the set measurements menu
> work for you?  I'm not sure what you mean by equivalent circular particles
> since you can obtain a radius from either the area or perimeter.  The
> documentation explains it better.
>
>           Or are you trying to segment the image into a collection of
> roundish features?  Voronoi or watershed may help.  I'm currently using a
> macro I wrote to identify non-interpenetrating circles to identify closely
> spaced features that are assumed to be round.
>
> John D.
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

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

[BenTupper]

Hi,

On Apr 17, 2013, at 6:49 PM, Adam Hughes wrote:

> Hey John,
>
> It's kind of hard to know which shape descriptors would be ideal.  I guess
> the best for me to do is to look at as many of the images as I can and see
> which shape descriptors may correspond best to the deviation I'm seeing.
> My deviations, for example, may be more suited to one correction over
> another.  I don't think I thought the question through before posting, and
> for that I apologize.
>

It might help if you start with an example we can all follow along with.  Here's a macro that anyone can run - that gives us all a common starting point.

run("Blobs (25K)");
setAutoThreshold("Default");
//run("Threshold...");
setThreshold(126, 255);
run("Convert to Mask");
run("Set Measurements...", "area centroid perimeter bounding fit shape feret's display redirect=None decimal=3");
run("Analyze Particles...", "size=0-Infinity circularity=0.00-1.00 show=[Bare Outlines] display clear record");

The outputs in the result table are defined here...

http://rsb.info.nih.gov/ij/docs/guide/146-30.html#toc-Subsection-30.7

Then there's a chance to get at the answer to John's question, "what you mean by equivalent circular particles?"

Cheers,
Ben

> On Wed, Apr 17, 2013 at 6:35 PM, John Dunsmuir <[hidden email]> wrote:
>
>> Adam,
>>           Do any of the shape descriptors in the set measurements menu
>> work for you?  I'm not sure what you mean by equivalent circular particles
>> since you can obtain a radius from either the area or perimeter.  The
>> documentation explains it better.
>>
>>          Or are you trying to segment the image into a collection of
>> roundish features?  Voronoi or watershed may help.  I'm currently using a
>> macro I wrote to identify non-interpenetrating circles to identify closely
>> spaced features that are assumed to be round.
>>
>> John D.
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

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