It’s not my style to cause offence, so I hope no one was put out by that title, but I wanted to draw your attention to the key issue in lighting. Consider this an expansion of my earlier posting, “Guidance on machine vision illumination.”
It’s absolutely critical when you determine how to illuminate your part or target, that you stop thinking like a human and start thinking like a camera. By this I mean you’re not trying to take a “nice” picture for the family album; you’re trying to enable a computer to separate out what it needs to detect in a image from the background “noise” that doesn’t matter.
This means that, in the ideal world, you’d make the thing you need the computer to recognize, (perhaps an edge that you intend to measure from,) black, and everything else white. Or vice-versa. The computer doesn’t care which way round you do it.
The best example of this is backlighting. When you backlight a part or target you make the background white and the object you’re interested in black (because it’s in shadow.) The resulting image will lack any artistry but will be easy to process and analyze.
Just something to bear in mind next time you’re wrestling with lighting.
Subscribe to:
Post Comments (Atom)
Posts
No comments:
Post a Comment