Posted by
Gabriel Landini on
May 26, 2010; 2:02pm
URL: http://imagej.273.s1.nabble.com/Analyze-particles-Help-and-bug-tp3688137p3688148.html
On Wednesday 26 May 2010 14:16:58 you wrote:
> But, the outputs differ and the outlined of the circularity set is not
> fully overlapped with the previous outline as generated in image #1.
> So the outcome looks distorted (image #4:
>
http://sites.google.com/site/tamjidulhoque/home/resume-of/useful-links/3_Ci> rcularitybasedsubsetMaskfirst.zip?attredirects=0&d=1) and for some
> images/case the outputs look very bad.
> So, please help on this issue. I am not sure if it is due to different
> line-thickness or different outlining algorithms ....
I think that this may be due to a known mismatch of the outlines of a ROI with
the pixels of a ROI. This was raised recently in the fiji-devel list.
I do not think that there is a solution currently unless you treat the blobs
in the same way.
My suggestion is to make your own routine to draw the boundaries of ROIs as
the pixels that touch the background. For example, assuming white objects on
a black background:
setBatchMode(true);
a=getTitle();
run("Duplicate...", "title=test-1");
selectWindow("test-1");
run("Minimum...", "radius=0.5");
imageCalculator("Subtract", a,"test-1");
setBatchMode(false);
You can do this also with Binary Thinning in a 1 line macro:
run("BinaryThin ", "kernel_a='2 1 2 1 1 1 2 1 2 ' rotations='none'
iterations=1 white");
This guarantees that all the pixels of the boundary of the blob are contained
in the blob that you see.
Otherwise, IJ gets the ROI of the blob and draws the outline which in not
completely contained in the blob.
If you want to show certain blobs, then use the XStart YStart coordinates to
produce an image with only those pixels set and use the BinaryRreconstruct
plugin in my page to recreate only those particles (the seed is the points you
set and the mask is the original binary image).
Cheers
G