I need trustworthy examples for our UR5e camera station. The same polished cover gains a bright stripe after reseating, and somebody labels the new image scratched. The physical cover has not changed. How do we stop teaching the inspection to recognise our setup variation?
Check the physical cover under the agreed inspection conditions first. Then capture the identified sample repeatedly with the lighting fixed. You'll have a much better question if you know which marks are really there.
Quality has accepted the cover. There is a small rub near one fixing hole, within their allowed example. The broad stripe does not correspond to a mark they can find.
Good, keep the rub in the image comparison rather than cropping it out to concentrate on the stripe. You need to see how the setup treats a known physical feature too.
Yes, but don't assume the corner explains every image difference. Keep the same camera and lighting settings through that comparison or you'll lose the useful link again.
Designer found contact at the rim where the trial nest was meant to leave clearance. They are revising that area. Camera and light settings are recorded for the before-and-after captures.
The revised nest seats that cover more consistently in the bench check. New images still have some bright variation, though the big change after reseating is reduced. I am not calling the inspection finished.
And have quality choose that reject by inspecting the part, not by finding the most dramatic picture in the old folder. You've just seen how persuasive the wrong bright line can be.