Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

colour to grey scale and crack detection

Solved!
Go to solution

One suggestion would be switching to a grayscale camera.  There is no reason for a color camera to be used in this application.  It would improve the details in the image.

 

Will real cracks look like this?  These look like a notch was cut in the rail with a radial saw.

 

Brighter lighting might improve the contrast between the rails and the cracks.  It looks like the cracks will be black no matter what, so with high enough contrast you could pick them out fairly easily.  Using a vertical edge detection matrix in the convolution vi would probably highlight the edges nicely.  Something like:

 

1 1 1 1 1 1 1

1 1 1 1 1 1 1

1 1 1 1 1 1 1

0 0 0 0 0 0 0

-1 -1 -1 -1 -1 -1 -1

-1 -1 -1 -1 -1 -1 -1

-1 -1 -1 -1 -1 -1 -1

 

This would highlight the top edge of the crack without highlighting much else.  Perhaps after that you could use a fixed threshold and see if any crack segments show up.

 

Bruce

Bruce Ammons
Ammons Engineering
0 Kudos
Message 11 of 15
(1,530 Views)

Hi sailesh,

 

I tried something on your images. You can optimize it further.

0 Kudos
Message 12 of 15
(1,502 Views)
Did the vi help?
0 Kudos
Message 13 of 15
(1,403 Views)
any update?
0 Kudos
Message 14 of 15
(1,337 Views)

I need to extract the crack alone in the image of a cracked egg, please kindly help me on this .

Download All
0 Kudos
Message 15 of 15
(710 Views)