Binary mixture and segmentation

S1_clean.tif (303.2 KB) S1_FeatJ_Edg_sub_bin_erx1_dilx1_OL.tif (909.4 KB) S1_FeatJ_edges.tif (1.2 MB) S1_marked.tif (909.4 KB)



Attached images are of a sample that contains particles, binder and free volume area (pores). I marked up an image for illustration.

Analysis goals

The main goal is to get the best segmentation to allow particle counter to isolate as many of the particles embedded in the adhesive. I want to retain as much of the original shape of the particle for size, area and shape factor analysis.


  • At this point I don’t think I’ve found the optimum work flow to capture as many of the particles in the original image. Used multiple schemes along with Auto Local Threshold (Phansalker r=15, P1=P2=0.5)

  • Have used the MorphoLibJ and Greyscale attribute filtering

  • Have used edge detection and FeatureJ to create an image which is then subtracted from original image (images attached as well as flattened overlay)

The schemes I mentioned are what I tried, but I’m reaching out to experts who probably know more about how to use these tools effectively.