Hello Benny -
I would suggest you start with:
Image > Adjust > Color Threshold...
Analyze > Analyze Particles...
I’m not exactly sure which colonies you want to include in your
two groups, but I think when you run
something like this might work:
Threshold on “Brightness” from 0 to 180 with “Pass” checked
(to screen out some of the background).
Then threshold on “Hue” from something like 135 to 160 with
“Pass” checked to get your “blue” colonies, and (still with “Hue”)
uncheck “Pass” and and threshold from something like 120 to
150 to get your “red/purple” colonies.
When you run
Analyze Particles... tightening “Circularity”
(to be close to one) will tend to exclude colonies that have merged
together, and loosening “Circularity” will tend to include them.
(Typically, you will have to experiment some with parameters and
algorithms to find an approach that works adequately well for your
The image you posted is JPEG, a lossy compression format.
Both for your own processing and for posting here you will be
better served to use non-lossy formats – you keep all of the
original information and avoid compression artifacts.
I don’t know whether the grid is part of the original image or
added later, but either way, if you can remove the grid, or
acquire the original image without the grid, it will simplify your
analysis. (In general, anything you can to to acquire cleaner,
better-quality images will make your life easier when it comes
time to process them.)
I assume that you want to analyze multiple images of this type.
It would be helpful you could post more that one such image
so we can get an idea of what kind of image-to-image variability
you are facing.
If you have a lot of images to analyze and you are hoping to
automate (some part of) the workflow, you should also start
playing around with scripting / macro programming.