Analyzing droplets in SEM-images

Hi All.
I am quite new to ImageJ and realize that there are countless of options avaliable, but where to start?
Maybe this question has been adressed already but:

  • In short : What I want to do is to count the number of metal droplets and their sizes in images like the attached two.
    Any hint about where to start would be appreciated,
    Wika Segmentation?

With best regards

Hey @ricols

Very cool images! What do you consider a ‘droplet’? Only the very circular objects?

Here are a few helpful links if you are new to ImageJ and Segmentation:

Just provide as much info as you can for what you aim to do for your analysis - so you only need their size and number?



Hi @etadobson
Thank you very much for the quick reply.
Yes, it is the only the round circular objects (the drops, or droplets) that are of interest.
If I could find some way to (at least at ~90%) identify the droplets, I then can calculate the number of drops /mm^2. If I could get the sizes as well that would be a bonus because the size distribution is also of interest in this case.
Below is a picture with the droplets of interest marked with white circles.

/ @ricols

Good day Rickard Olsson,

thanks for the details.

To isolate what you’ve indicated as droplets will be rather difficult. Size invariant correlation comes to my mind but if you consider the following excerpt, you will realize that it is difficult to decide by eye what is a droplet.

Furthermore, your sample image is defective: There is a break at about 57pel from the top. Is this image from literature?




Check out the links above… it’s true it won’t be easy and might not be successful in fully automating this segmentation… Other experts on here might chime in on possible methods. A great tool to also look into would be @iarganda’s MorphoLibJ.

Will test a few things out myself to see if I can help you and will post if I come up with anything… in the meantime - go through all those links to shore up on options in ImageJ/Fiji.

eta :slight_smile:

1 Like


Just to give you a bit of hope… Here are the inital results I got with quickly playing with MorphoLibJ’s Morphological Segmentation tool:

It’s not ideal yet - there is room for improvement… but you can get there! :slight_smile: Then you can just select your more circular objects and off you go!


1 Like

Hi @anon96376101
Thank’s for the feedback. No the SEM-imnages are taken by me and yes in this one there is a jump at the line you indicate probable because I changed scanspeed or something.
Also, yes, it is not 100% clear what is a droplet, but at least to identfy the obvious ones would be of great help because then I can get a correlation of droplet coverage percentage vs. production parameters.

Hi @etadobson
Thank you very much for your efforts and sorry for my late reply. I haven’t had any time this weekend for this .

I’ll have a look at it again today though.
MorphoLihJ you say.

1 Like