Wednesday, October 29, 2008

Calibrating calibration grids

A recent posting on the LabVIEW LAVA board asked how to deal with an ISO auditor who wanted chrome-on-glass grids calibrated.

Let’s discuss.

First, an explanation for the uninitiated: If you’re doing gauging with machine vision you need to convert pixels to real world measurements. That entails creating a conversion factor, the precision of which will establish the best practicable performance of your vision system. (For simplicity we will overlook the situation where you calibrate to remove optical distortion.)

You get the conversion factor by finding out how many pixels span a known distance. The best way to do this is by using a precision array of dots on a glass surface, such as those manufactured by
Max Levy. (Making sure that the glass target is at the same working distance as your regular target!)

So, assuming you have the good sense to check the calibration of your vision system regularly, how often do you need to check the target?

Well given that it’s not subject to any wear, I will argue that all you need do is check it’s undamaged. No recalibration is required. But those ISO guys can be a special breed, so if you can’t win that argument you’ll need to get your glass grid recalibrated. For that I suggest you contact
Max Levy. (Yep, them again.)

And if you do have your grid checked for accuracy, please let me know if it had changed.

No comments: