Alpha shape macro available

[Jonathan Lake]

Hello all,

I have written a ImageJ1 macro that computes the alpha shape for a given point selection. It works analogously to the “Convex Hull” command but can produce concave and hollow shapes depending on the particular point set and alpha value. The output is a composite selection of the alpha shape. Give it a try and let me know if it works for you.

Three known issues:

1 - Technically the alpha shape can contain isolated edges that are not part of a polygon. These are not calculated by this script because ImageJ lacks the ability to combine multiple line ROIs. Generally these isolated edges are not practically useful.
2 - The script gets slow and can take a few minutes to complete once the number of points exceeds 200 or so. Constructing the Delaunay triangulation is slow.
3 - The script uses the ROI manager and will clobber anything that is in there when it is called.

Script download: http://jonlake.com/download/Alpha_Shape.ijm

A description of alpha shapes: http://cgm.cs.mcgill.ca/~godfried/teaching/projects97/belair/alpha.html
--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html

[Gabriel Landini]

On Sunday 13 Jul 2014 14:32:42 Jonathan Lake wrote:
> I have written a ImageJ1 macro that computes the alpha shape for a given
> point selection. It works analogously to the “Convex Hull” command but can
> produce concave and hollow shapes depending on the particular point set and
> alpha value. The output is a composite selection of the alpha shape. Give
> it a try and let me know if it works for you.

Thanks for posting this, but unfortunately it does not work under in my
machine linux and openjdk 1.7.0. I am updated to the latest daily build.
I create a new 8bit image, create a multi-point selection and run the macro,
but then it shows an empty dialog and then it freezes.

Cheers

Gabriel

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

[Jonathan Lake]

Hmmm. Unfortunately I cannot readily replicate your environment to test it under those circumstances. It was written in Fiji with ImageJ 1.49d on OS X Mavericks. I have no idea what could cause a problem before it even manages to display the dialog.

try this version?
https://www.dropbox.com/s/rvy2zsyb2cwbsus/Alpha_Shape_alternative.ijm

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

[Gabriel Landini]

On Sunday 13 Jul 2014 16:59:35 Jonathan Lake wrote:
> Hmmm. Unfortunately I cannot readily replicate your environment to test it
> under those circumstances. It was written in Fiji with ImageJ 1.49d on OS X
> Mavericks. I have no idea what could cause a problem before it even manages
> to display the dialog.
>
> try this version?
> https://www.dropbox.com/s/rvy2zsyb2cwbsus/Alpha_Shape_alternative.ijm

Tried it on Oracle Java 1.6.0_35. Similar story:
It now shows the dialog, asks for a number (I leave the default of 0). Then it
given an error saying:
-----------------
Empty array in line 655.

currentpoly - newArray ( uniquealphaedgelist [ currentpoint <]> ) ;
-----------------

This is the same in IJ1 as in Fiji on Oracle Java.

Cheers

Gabriel

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

[karo03]

Hi Gabriel,
on my system it works with alpha given with a comma: e.g. 5,0 not 5 or 0,0 , but i think the result is always the convex hull. Seemingly there is some number type error!

Regards
Karsten

Am 14.07.2014 um 12:56 schrieb Gabriel Landini <[hidden email]>:

> On Sunday 13 Jul 2014 16:59:35 Jonathan Lake wrote:
>> Hmmm. Unfortunately I cannot readily replicate your environment to test it
>> under those circumstances. It was written in Fiji with ImageJ 1.49d on OS X
>> Mavericks. I have no idea what could cause a problem before it even manages
>> to display the dialog.
>>
>> try this version?
>> https://www.dropbox.com/s/rvy2zsyb2cwbsus/Alpha_Shape_alternative.ijm
>
> Tried it on Oracle Java 1.6.0_35. Similar story:
> It now shows the dialog, asks for a number (I leave the default of 0). Then it
> given an error saying:
> -----------------
> Empty array in line 655.
>
> currentpoly - newArray ( uniquealphaedgelist [ currentpoint <]> ) ;
> -----------------
>
> This is the same in IJ1 as in Fiji on Oracle Java.
>
> Cheers
>
> Gabriel
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

... [show rest of quote]

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

[Gabriel Landini]

On Monday 14 Jul 2014 13:17:54 Karsten wrote:
> on my system it works with alpha given with a comma: e.g. 5,0 not 5 or 0,0 ,
> but i think the result is always the convex hull. Seemingly there is some
> number type error!

Oh, yes I can reproduce that behaviour.
Odd, because the comma should not be the decimal separator.
I also get the convex hull (and all the points disappear).

Thank you

Gabriel

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

[karo03]

.. very good, at least the same behaviour.
I think there is a problem with Dialog.getNumber().
Still I have to leave now!

Karsten

Am 14.07.2014 um 13:26 schrieb Gabriel Landini <[hidden email]>:

> On Monday 14 Jul 2014 13:17:54 Karsten wrote:
>> on my system it works with alpha given with a comma: e.g. 5,0 not 5 or 0,0 ,
>> but i think the result is always the convex hull. Seemingly there is some
>> number type error!
>
> Oh, yes I can reproduce that behaviour.
> Odd, because the comma should not be the decimal separator.
> I also get the convex hull (and all the points disappear).
>
> Thank you
>
> Gabriel
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

... [show rest of quote]

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

[karo03]

perhaps ImageJ Macro does not properly digest that mixing between main code and blocked subfunction code!

Still I have to leave, really, now!

Karsten

Am 14.07.2014 um 13:26 schrieb Gabriel Landini <[hidden email]>:

> On Monday 14 Jul 2014 13:17:54 Karsten wrote:
>> on my system it works with alpha given with a comma: e.g. 5,0 not 5 or 0,0 ,
>> but i think the result is always the convex hull. Seemingly there is some
>> number type error!
>
> Oh, yes I can reproduce that behaviour.
> Odd, because the comma should not be the decimal separator.
> I also get the convex hull (and all the points disappear).
>
> Thank you
>
> Gabriel
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html

... [show rest of quote]

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

[Thomas Boudier]

Hi,


After different attempts I could run it, and it is working fine, depending
on the value you get convex hull or proper a-shape. I did notice that the
points must be quite close one to another. Hope this helps.

Thomas

> perhaps ImageJ Macro does not properly digest that mixing between main
> code and blocked subfunction code!
>
> Still I have to leave, really, now!
>
> Karsten
>
> Am 14.07.2014 um 13:26 schrieb Gabriel Landini <[hidden email]>:
>
>> On Monday 14 Jul 2014 13:17:54 Karsten wrote:
>>> on my system it works with alpha given with a comma: e.g. 5,0 not 5 or
>>> 0,0 ,
>>> but i think the result is always the convex hull. Seemingly there is
>>> some
>>> number type error!
>>
>> Oh, yes I can reproduce that behaviour.
>> Odd, because the comma should not be the decimal separator.
>> I also get the convex hull (and all the points disappear).
>>
>> Thank you
>>
>> Gabriel
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
>

... [show rest of quote]

--
Thomas Boudier, Associate Prof. UPMC,
IPAL UMI 2955, Singapore.

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