Home Image-Pro General Discussions

Droplet count from a surface

Hi all,
I need a little bit help from the experts here. I wanted to count number of droplets in this surface with the diameter (mean diameter). I tried using smart segmentation but because some of my droplet has similar intensity as the background I am unable to segment it perfectly. Same for manual segmentation. Is there any better way to do this?
Thanks in advance.

Answers

  • 2023-02-03-140612

    raihanul --

    It is not perfect but here is a method that is pretty quick.
    1. SET A BRIGHT THRESHOLD TO FIND THE REFLECTION ON EACH DROPLET
    2. COUNT THE OBJECTS IDENTIFIED BY THE BRIGHT THRESHOLD
      (SEE AA IN FIRST SCREEN CAPTURE BELOW)
    3. CREATE A VORONOI MAP OF THE OBEJCTS IDENTIFIED IN 2
      (SEE BB IN FIRST SCREEN CAPTURE BELOW)
    4. TRANSFER THE VORONOI MAP ONTO THE ORIGINAL IMAGE
      (SEE CC IN FIRST SCREEN CAPTURE BELOW)
    5. MEASURE DROPLET AREA BY MEASURING VORONOI REGIONS
      (SEE SECOND SCREEN CAPTURE BELOW)
    These results may be improved by doing a FLATTEN on the IMAGE before STEP 1.

    I hope this information is helpful.

    -- Matt






  • Dear Matt,
    thanks for showing me a way to solve this particular problem. I tried in several different ways. So after doing some 2D filtering I actually tried to perform manual segmentation using the thresholding tool and some parameter changes in the option of count tab. The best result I could get is shown below. But still there is some problem in thresholding. You can see in the attached image here there are portions of the large droplet in top left corner which are not considered (probably because of the dark shades on the right of each large droplets). Do you happen to know any solution I can use as for example any 2D filter to reduce the extreme dark shades but not changing the surface shades.

  • 2023-02-03-143839

    raihanul --

    You have a challenging image.

    Part of your DROPLETS are BRIGHT and part of them are DARK, and the fraction of BRIGHT and DARK varies across the image.

    This is shown below with the THRESHOLD TOOL set to MASK both the BRIGHT (160-255) and DARK (0-80) PIXELS.

    You may also want to investigate FLATTENING.  There are significant differences between the LEFT SIDE and the RIGHT SIDE of your IMAGE.  This is shown below with AA being your ORIGINAL, DD being after FLATTENING, and EE being the ABSOLUTE VALUE DIFFERENCE IMAGE generated by AA and DD.

    I hope this information is helpful.

    -- Matt



  • Dear Matt,
    Thanks a lot for showing me some way. I will try the way you showed in last comment.

Sign In or Register to comment.