Re: Wand thickness for Part. Analysis

Posted by Gabriel Landini on Jan 15, 2010; 11:59am
URL: http://imagej.273.s1.nabble.com/Wand-thickness-for-Part-Analysis-tp3689716p3689718.html

On 15 Jan 2010, at 03:06, Rothman, William wrote:
> I want to join particles (clumps) that are within 10 (or some other
> value)  pixels using particle analysis. I thought this could be
> done by changing the wand size. Is this correct and could someone
> describe how. I've tried to no effect.Would this be related to the
> Feret distance or Eucl.Dist???

If you want to join blobs, then binary dilation of the image with a disk of
diameter 5 or so should do (i.e. filter maximum with a radius of 5), but
obviously this changes the particles shape and size. Then you could use the
joined cluster -if it can be identified somehow among other possible clusters-
to binary_reconstruct the original image to obtain just the original objects
that were associated with the cluster.
This goes under the umbrella of spatial reasoning. Dave Randell, who is the
author of the renowned RCC8 spatial reasoning logic commonly used in AI, gave
a talk precisely on this at the 2008 Luxembourg meeting.
The abstracts are at the conference site http://imagejconf.tudor.lu
but I am not sure where the links to the full papers are in that site.

His paper on spatial logic (cited more than 1000 times!) is here:
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.35.7809&rep=rep1&type=pdf
but I do not seem to be able to open it right now.

I hope it helps.
Cheers
G.