different version of skeletonize produce different results

[Aryeh Weiss]

I have two skeletonize plugins that come with Fiji. The entries in the
command finder are:

Process>Binary    Skeletonize ij.plugin.filter.Binary("skel")

and

Plugins>Skeleton    Skeletonize (2D/3D)
sc.fiji.skeletonize3D.Skeletonize3D_
/Applications/local/fiji/Fiji.app/plugins/Skeletonize3D_-2.0.0.jar


I find  that these produce two different results. I attached a test
image and the two results
(skeletonizBinary is the first one, and skeletonize2D3D is the second).

Searching the archives for skeletonize turns up a number of posts that
indicate that there may be different valid ways to skeletonize (eg,
4-connected vs 8-connected).. So I wonder is there are methodological
differences between these two, an if so, what they are.

Thanks in advance
--aryeh

--
Aryeh Weiss
Faculty of Engineering
Bar Ilan University
Ramat Gan 52900 Israel

Ph:  972-3-5317638
FAX: [phone]


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

[Ignacio Arganda-Carreras-2]

Hello Aryeh,

The information about Skeletonize3D is here: http://fiji.sc/Skeletonize3D

It is an implementation of the thinning algorithm by Lee et al. (1994):
http://dl.acm.org/citation.cfm?id=202862.202867

There quite a few different skeletonization algorithms out there ;)

ignacio

On Thu, Mar 3, 2016 at 2:20 PM, Aryeh Weiss <[hidden email]> wrote:

> I have two skeletonize plugins that come with Fiji. The entries in the
> command finder are:
>
> Process>Binary    Skeletonize ij.plugin.filter.Binary("skel")
>
> and
>
> Plugins>Skeleton    Skeletonize (2D/3D)
> sc.fiji.skeletonize3D.Skeletonize3D_
> /Applications/local/fiji/Fiji.app/plugins/Skeletonize3D_-2.0.0.jar
>
>
> I find  that these produce two different results. I attached a test image
> and the two results
> (skeletonizBinary is the first one, and skeletonize2D3D is the second).
>
> Searching the archives for skeletonize turns up a number of posts that
> indicate that there may be different valid ways to skeletonize (eg,
> 4-connected vs 8-connected).. So I wonder is there are methodological
> differences between these two, an if so, what they are.
>
> Thanks in advance
> --aryeh
>
> --
> Aryeh Weiss
> Faculty of Engineering
> Bar Ilan University
> Ramat Gan 52900 Israel
>
> Ph:  972-3-5317638
> FAX: [phone]
>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
Ignacio Arganda-Carreras, Ph.D.
Ikerbasque Research Fellow
Departamento de Ciencias de la Computacion e Inteligencia Artificial
Facultad de Informatica, Universidad del Pais Vasco
Paseo de Manuel Lardizabal, 1
20018 Donostia-San Sebastian
Guipuzcoa, Spain

Phone : [phone]
Website: http://sites.google.com/site/iargandacarreras/
<http://biocomp.cnb.csic.es/~iarganda/index_EN.html>

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

[Aryeh Weiss]

Hi Ignacio

Thanks for your quick reply.

On 03/03/2016 3:49 PM, Ignacio Arganda-Carreras wrote:
> Hello Aryeh,
>
> The information about Skeletonize3D is here: http://fiji.sc/Skeletonize3D
>
> It is an implementation of the thinning algorithm by Lee et al.
> (1994): http://dl.acm.org/citation.cfm?id=202862.202867
>
> There quite a few different skeletonization algorithms out there ;)
>

What interests me is a way to describe intuitively what the difference
is between the two methods.

In comparing these on my test image, I have the initial impression that
the Process>Binary>Skeletonize version produces a super set of the
branches produced by Skeltonize2D3D. I attached a composite with the
Skeletonize2D3D in red. The object's edge is in the gray channel.

It is as if one version "smooths" surface variations to avoid having
every pixel become an endpoint.
I dont think that this is explicitly done, but rather it may be implicit
to the algorithm that is used.
This reminds me of the idea that watershed splitting has to be limited
to avoid over segmentation.

Best regards
--aryeh

> ignacio
>
> On Thu, Mar 3, 2016 at 2:20 PM, Aryeh Weiss <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>     I have two skeletonize plugins that come with Fiji. The entries in
>     the command finder are:
>
>     Process>Binary    Skeletonize ij.plugin.filter.Binary("skel")
>
>     and
>
>     Plugins>Skeleton    Skeletonize (2D/3D)
>     sc.fiji.skeletonize3D.Skeletonize3D_
>     /Applications/local/fiji/Fiji.app/plugins/Skeletonize3D_-2.0.0.jar
>
>
>     I find  that these produce two different results. I attached a
>     test image and the two results
>     (skeletonizBinary is the first one, and skeletonize2D3D is the
>     second).
>
>     Searching the archives for skeletonize turns up a number of posts
>     that indicate that there may be different valid ways to
>     skeletonize (eg, 4-connected vs 8-connected).. So I wonder is
>     there are methodological differences between these two, an if so,
>     what they are.
>
>     Thanks in advance
>     --aryeh
>
>     --
>     Aryeh Weiss
>     Faculty of Engineering
>     Bar Ilan University
>     Ramat Gan 52900 Israel
>
>     Ph:  972-3-5317638
>     FAX: [phone]
>
>
>     --
>     ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>
>
>
>
> --
> Ignacio Arganda-Carreras, Ph.D.
> Ikerbasque Research Fellow
> Departamento de Ciencias de la Computacion e Inteligencia Artificial
> Facultad de Informatica, Universidad del Pais Vasco
> Paseo de Manuel Lardizabal, 1
> 20018 Donostia-San Sebastian
> Guipuzcoa, Spain
>
> Phone : [phone]
> Website: http://sites.google.com/site/iargandacarreras/ 
> <http://biocomp.cnb.csic.es/%7Eiarganda/index_EN.html>

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