Detect defects on Al fracture surface

Hi Everyone,

I am new to imageJ and want to analyze the fracture surface of a 3D printed Aluminum. The fracture surface was imaged using SEM (secondary electron mode), Jimage_fail20191016.000005.tif (1.2 MB), (288 pixels equals 1 mm). As can be seen in the attached file, the fracture surface contains many defects. I want to know the area covered by these defects as to compare them for different samples. In my first attempts I used the following procedure:

  • Apply Bypass filter
  • Threshold the image
  • Analyze particles

I was able to get a percentage of defects covering the surface area. However, the results are not reproducible and deviate quite a bit. I reckon I need to apply more filters before Thresholding, unfortunately I do not know which to use.

Any tips would be very much appriciated :slight_smile:!



So what do you consider the defects on the surface? The darker circles?

I gave it a go just directly using Trainable Weka Segmentation (TWS) plugin that comes with Fiji. Just using the default parameters and drawing a few traces - I got this:

You can also go the more ‘classic’ way as you were… but knowing a bit more about what you consider ‘defects’ will help us better help you improve your current workflow. But TWS is also an option. :slight_smile:

1 Like


Thank you for your response!

Indeed, the dark circles are the defects. Some of them are surrounded by large "dimples"as shown in the figure below ;

I would like to separate only the dark circles (defects), so not the surrounding “dimple”