Detecting particle thickness and eliminating those particles with thickniss less than for a given value

[alexderjuengere]

hi desalando,

as an alternative to eroding pixels, you might convolute your 2-color input with an 2d gauss-kernel and then calculate the global maximum in each row or column.

[BruceAmmons]

Yes, I knew you would get the same results from odd and even widths.  I was going to mention that, but I forgot.  When you divide by two and get an integer, it is kind of implied.  You can't get half a pixel.

I suppose one solution would be doubling the resolution of your image, then running it.  The results would be correct for the original image, I believe.  It would be an interesting experiment.

Bruce