Hi,
it looks difficult.
As a starting point, to make manual selection easier, you could try
the Versatile Wand
http://imagejdocu.tudor.lu/doku.php?
id=plugin:segmentation:versatile_wand:start
It has preview, so you can vary the value tolerance and also the
"color" sensitivity to get the best result.
Then you can modify the selection manually with the freehand
selection tool (use shift/alt to add/remove areas).
If this results in a selection with several separate areas, make a
mask (Edit>Selection>Create Mask) and run "Analyze Particles" on the
mask.
If your original image has very good resolution and low noise (the
one on mypicx is rather poor quality), you can also try whether the
variance differs anough between the areas with stationary and falling
sand to make a distinction (maybe use some edge-conserving blur like
Thresholded Blur after the Process>Filters>Variance)
http://imagejdocu.tudor.lu/doku.php?
id=plugin:filter:thresholded_blur:start
Michael
________________________________________________________________
On 23 Nov 2009, at 05:20, Olukayode Ajayi wrote:
> Hi,
> Here is sample of my imageJ traced in imageJ manually(
>
http://www.mypicx.com/11222009/image_one/ ) and the results are
> computed to
> determine the area within the traced here.
>
> There are issues around the manual tracing such as time constraint
> when
> procesing large number of images and inconsistency in result due to
> manual
> tracing of region of intereste.
>
> I could have developed a macro based on the first image traced but the
> location of the manually traced image changes as the drum rotates,
> which
> makes it impossible to do.
>
> I'll appreciate any advice on how to do this without manual tracing
> in imageJ.
>
> I look forward to your reply
Locked
