Wednesday, March 9, 2011

Vision Troubleshooting #4

We know the product being inspected hasn’t changed, we know the system used to run much better than it does now, and we’ve attended to all the basic cleaning tasks to ensure the light intensity is correct, the background is clean, and so on. Now what?

Here’s a back-to-basics checklist:

  1. Is everything powered up and working? Are the lights, camera and PC all on? (This like asking if there’s gas in the tank. Yes, it’s basic, maybe even insulting, but check anyway. You wouldn’t be the first person to find that the lights were all out.)
  2. Is the system actually acquiring an image? This will tell you whether the cameras are being triggered and are in communication with the PC.
  3. Is the product under inspection in the right place when the image is grabbed? It’s not uncommon to find a sensor has moved just a few millimeters or a conveyor is running slightly slower than it used to.
  4. Are the vision tools finding the part? Most applications start with some kind of part location – perhaps a pattern match or maybe a simple edge detection. Step through the inspection (you might need to take a set of images and put them through an emulator,) and verify that the edge or trained pattern is actually being found. If not, you’ve probably got some kind of part or lighting change.
  5. Look again at the part. It’s been my experience on many occasions that a sudden change in vision system performance actually means there’s been a sudden change in the part the vision system is looking at.
And if those tips haven’t brought system performance back to where it once was, well you could try dropping me an email. I might have some ideas.

Well that’s almost the end of my MV4U Vision Troubleshooting tips; there’s just one thing left to address: the vision don’ts. That’s Vision Troubleshooting #5.

No comments: