Vision system to look at a bubble level?

[jeff_scharpf]

Would it be possible to use some type of NI vision system to view a bubble in a bubble type level and to somehow "measure" how far off the bubble is left or right?

This would be some type of system that would (somehow) set a table to a calibrated "known" level position. I'm not sure if this would be some type of gyroscope or another calbirated level, but we would have to start with a good solid calibrated level surface. We would then put a level to be tested and use a vision system to "view" the bubble to see how far off it is and report it as a % accuracy or uncertainty.

I'm not looking for Labview code, only if this can be done with some NI vision system. Is there a camera that can see the bubble and calculate some % of the bubble outside of the lines?

Seems like an odd application I know, but I'm just looking for input.

[ogk.nz]

There are a lot of different variables here, like camera orientation, type of level, coloration of liquid, background etc. 

However it seems as if this should be reasonably doable, if you mount the camera to the platform so that it is in a fixed position with relation to the centre, and you calibrate the program you should be able to reasonably accurately determine the position of the table. I would imagine with a lot more accuracy then using your eye on the same system.

I'm not entirely sure if you would be able to do it with just an NI smart programmed but it seems doable. However it is entirely possible using just any webcam and a PC on the side if you are only using this in a lab for testing/calibration.

[jeff_scharpf]

Thank you for the response. Yes I understand that there are a lot of variables.Currently, this is done manually, with a person looking at it. If an automated, or semi-automated method is found, we may use this as a sort of "end of line" check. I am not yet aware of the different types or colors. I had not thought of that. These are good points.

[AnalogKid2DigitalMan]

Alternative approach I'd use if space permits and tilt angle is within reason:

Use a digital protractor and then use vision tools for character recognition to read the display. If angle range is high, then use a fiducial and coordinate system to compensate for the angle so the characters can still be read.

http://www.globalindustrial.com/p/tools/test-measurement/Measuring-Protractors/pro-360-digital-protr...

EDIT: There are also inclinometers that have analog and digital ouputs/communications that would alleviate the need for vision entirely.

-AK2DM

[jeff_scharpf]

We definitely will need a reference such as that, however, once we know what the table is set to, we will need to see a level on the table and compare the level bubble reading with what we "know" to be the true level. So yes for the table or "standard" I think we can use a calibrated level with some type of output if possible, and this would get us to where we know what level the table is.

However, since we're trying to verify the bubble in a level against this known standard, so I think we still need a vision system to "see" the bubble.

Not sure if this makes sense.