How light shapes image data in machine vision

Occlusions = when the obvious gets overlooked

When you place a camera above an object, it seems natural to assume you’re looking at it from the top.

Sounds obvious, right?

But here’s the twist: I really find it hard to understand that there are situations, especially when using wide-angle lenses, when you see the object almost from the side, even though the camera is positioned directly above the object.

You say, sure that is obvious?

But I’ve found, time and again, that this detail gets underestimated.
Not just by beginners, but also by experienced engineers and customers alike.

How foreground objects block what cameras can see

The moment you have something in the foreground, closer to the lens than your actual object of interest,
that front object can block the view to your target.

That’s what we call an occlusion.

How to detect occlusions early and prevent costly redesigns

It might seem like a minor oversight, but it can become a real showstopper. Especially when the setup has already been calculated, components are ordered, and you discover too late that the camera “sees” something completely different from what you had planned.

Takeaway: Perspective matters

Even if your camera is in the “right” place on paper, ask yourself: What does it actually see?
Walk through the viewing cone. Simulate or test early.

Because once occlusion happens, no software can fix what the camera never saw.

When was the last time your camera “saw” something completely different than you expected?
Have you ever had an occlusion derail a project plan? How did you solve it?

Mail to Patrick