Revision history [back]

I have the original image, my goal is to identify and count the particles exactly for which I am using blob detection (as they may not be spherical sometimes and also houghs didn't work that well). I tried threshold the image with some fixed value but it misses some fainter particles, so I tried local OSTU by dividing the image in 10x10 grid and I got the second image but the problem is that the particles get fragmented i.e. see the top particle on the right side in the thresholded image, it is actually a single thing but now appears as multiple small particles, and when I use contour for particle count I get incorrect image. Also, some particles appear to be merged are not counted properly.

I have the original image, my goal is to identify and count the particles exactly for which I am using blob detection (as they may not be spherical sometimes and also houghs didn't work that well). I tried threshold the image with some fixed value but it misses some fainter particles, so I tried local OSTU by dividing the image in 10x10 grid and I got the second image but the problem is that the particles get fragmented i.e. see the top particle on the right side in the thresholded image, it is actually a single thing but now appears as multiple small particles, and when I use contour for particle count I get incorrect image. Also, some particles appear to be merged are not counted properly.