Creating binary image from SEM image of ductile dimples on fracture surface

@Ilya_Belevich,
The BM3DF filter is amazing! Thanks!
I found that the BM3DFSHARP sharpens the image as well.
I will read about the marker controlled watershed.
Thank you,
Florin

Well Hello again Florin,
Maybe you have already found an answer, as I said ‘expert in no time’.
As you can tell from the image, the illumination source (electron gun), the target, and the aquisition point are all angled into a triangle which doesn’t make the image ideal but then you will seldom work with ideal images.
Also you will note that these are not ‘particles’ as such but more like an image of a mountain range seen from space so if you don’t mind I will speak (type) to that example.
The data you obtain from this image will be best described as altitude of the mountain, depth of the valleys, length of ridges etc…
Being the energy conservationist I am (lazy), the best way to start will be to go to Plugins > Segmentation > k-means Clustering. I would start with no more than 4 clusters, and if you want more you can always increase. Check ‘Show clusters as centroid value’, Print optimization trace and Send to results table. This will show your image in 4 values to segment from in the table. The printed images that will be provided can have a LUT applied to them, use any choice you prefer.
Your segmented image can be binarized and analysis applied to it for the data you wish.
If this does not provide the data you require just ask more, there are many ways to do what you want. And hang on to the 3D image stack, it could be a valuable tool for your analysis
Good luck,
Bob

1 Like

Thanks Bob!
Interesting way of doing it. I will try it next week.
It isn’t clear in your message but I gathered that you are talking about the IJ-Plugins Toolkit for ImageJ. I installed it and played a little but with the 3D Viewer. I hope there is some documentation somewhere so that I can get to understand what is going on.
Regards,
Florin