Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

effective particle separation

Hi, all

I came across a problem of separating particles connecting together in a image. i attached the figures and diagram. Particle A, B, C and D should separate particles, and such is the case for other particles. i checked the forum, there is one thread said that the Gaussian or Laplacian filters can be used in this case. see:

http://forums.ni.com/t5/Machine-Vision/separation-of-particles/m-p/322071

But i tried several times, i still could not get the ideal result.

I am not sure whether it is ok to use user-defined filters. Because i don't whether the area of these particles after filtering will change or not.

cheers

Download All
0 Kudos
Message 1 of 5
(2,953 Views)

I think you may want to try using the Watershed algorithm. See this page for some advice on how to get started:

http://zone.ni.com/reference/en-XX/help/370281M-01/nivisionlvbasics/improve_the_binary_image/

0 Kudos
Message 2 of 5
(2,932 Views)

Hi, thank you very much for your reply. it is very helpful. however, my problem is still there. the separation efficiency is not so good.

i prefer to use the IMAQ separation, but the proper number of erosion is hard to get.

0 Kudos
Message 3 of 5
(2,926 Views)

It looks like with the amount of overlap between particles, you may not be able to separate the particles without significantly changing the area of the particles.

Zach C.
Field Engineer
Greater Los Angeles

0 Kudos
Message 4 of 5
(2,907 Views)

i checked that it is not possible to effectively separate the particles. alternatively, i used the mathematical model in this case. this is out of the scope of vision.

0 Kudos
Message 5 of 5
(2,891 Views)