Posted by
Michael Schmid on
URL: http://imagej.273.s1.nabble.com/Particle-Analysis-tp3690320p3690323.html
Hi Florian,
a few more ideas:
- use Process>Subtract background
- try the Versatile Wand from the documentation wiki with a gradient
threshold. You could click on the background, then invert the
selection, so you have all particles.
http://imagejdocu.tudor.lu/doku.php?
id=plugin:segmentation:versatile_wand:start
- if the background is rough everywhere, run
Process>Filters>Variance, then do some smoothing (maybe edge-
preserving, e.g. Sigma Blur from the ImageJ site or Thresholded Blur
from the documentation wiki). Maybe you can threshold this image more
easily.
Michael
________________________________________________________________
On 26 Nov 2009, at 12:17, Florian Sedlmeir wrote:
> Hi,
>
> I want to analyse particle-like structures with ImageJ. The problem
> is, that these particles usually have a good contrast against their
> local environment, but globally there are too many areas with
> similar or same grey values to threshold the image. So i have a few
> questions concerning this:
>
> - Is there an easy way to threshold only selected areas in a
> picture (to threshold the whole picture step by step)?
>
> - My particles are easily distinguishable with the eye because they
> are not as rough as their environment (+usually there is a local
> difference in contrast). ImageJ is able to find the edges with the
> "find edges" tool, but unfortunatly it doesn´t create closed areas
> (my plan was to find edges, threshold the resulting picture and
> analyse the resulting particles then) Is there a solution for this
> a problem?
>
> - One of my friends showed me the "magic wand" tool in Photoshop,
> which is able to determine my particles properly. Unfortunatly the
> same tool in ImageJ is not able to do this, it seems to be too
> sensitive. (Are there parameters which can be adjustet?)
> So my question is: Does a plugin exist, which is able to do the
> same task as Photoshop´s magic wand?
>
>
> Thank you and best regards.
> Florian