Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

2 Cameras 1 Measurement

Currently working with labview vision devolopment module and a Basler acA3800-10gm, our current fov is netting around 0.004"/pixel. we would like to be closer to 0.001" per pixel. Our target part is approximately 3"x12" and we are measuring the nominal dimensions along with other features.

 

To get the resolution down to 0.001" we are looking at using multiple cameras to capture images across the part. To get the nominal dimension we will need to take measurement data from 2 of the cameras. Is there built in feature for this within the vision devolopment module? is creating an overlapping coordinate system from a calibration the best bet then do the the arithmitic outside the VB processing?

0 Kudos
Message 1 of 2
(2,517 Views)

Hello,

 

It would likely be best to have a constant background that you can use as reference (for example, a ruler directly beneath/above the object). From here you can use Feature Extraction to identify and calibrate the two images. You can also do this without a constant background reference if your part has distinct features that you can use feature extraction. As you mentioned you will have to do some arithmetic outside of VB.

 

Some things to consider:

What are you looking to inspect? Does it have distinct features in the overlap section?

Have you considered moving to a linescan camera? These will be more capable of taking wider images though you will likely need some sort of stage to either move the part or camera.

 

Regards,

Sam K

 

0 Kudos
Message 2 of 2
(2,464 Views)