What Type of Structured Light to Use to Detect Surface Features on Highly Reflective surface

[BeardedBaron]

Hello,

I'm new to machine vision, and I'm trying to detect the surface features of a stack of highly reflective metal sheets in a low light environment.

The issue I'm running into is that there is not enough light to detect the features without using an external light source, but  normal camera flash creates too much glare for the surface features to be seen. 

Is there some specific light source I should be considering?

And in case it matters the camera I'm using can only detect in visible light.

Thank you.

[BruceAmmons]

You need to explain what surface features you are trying to detect.  That will make a difference.

You could use dark field lighting, which lights the part at a low angle.  The part will appear black, except for any features that reflect light into the camera.  This works well for scratches and rough spots.

You could use a cloudy day illuminator that lights from all angles.  This works well when you are trying to read the print on a shiny surface, or for other details that will show up in this type of light.

If neither of these would work, you will need to explain what features you are trying to detect for more help.

Bruce

[b.ploetzeneder]

Depends on how bad your glare situation / reflectivity / ambient light is. Sometimes just using a diffuser can help. Simply place some cloth / a white sheet in front of your light source and see if it helps.

Polarized light + pol filters usually work quite well on metal surfaces to reduce the influence of ambient light.. You can test that with cheap filters used for photography and then implement it with proper machine vision filters + foil to cover your light source.

If your surface is flat, use an enclosure to remove ambient light and then well-angle light. A simple concept is angling a flat light at 45 degrees + the camera at -45 degrees to get an almost straight angle. More expensive are properly collimated light sources / coaxial illumination. 

[BeardedBaron]

Hi Bruce,

I misspoke in my original post, I should have been more explicit, I'm trying to detect the edge between the thin layers of the stack of metal (the gaps between each sheet) not the actual surface itself. Is there anything I should be doing for that?

Below is an example of the style of images I'm trying to take.  The real sheets are much larger, and the gaps between the sheets might be much closer together or potentially even farther apart.

[BruceAmmons]

Ah, a totally different problem.  In this case, I would probably use a ring light.  You would get a pretty good reflection from the metal surfaces, with dark gaps between the plates.  If the size of the plates is too large, you can put a large fluorescent light on either side of the camera.  Experiment with top/bottom or side/side.  For ideal lighting, the light needs to be larger than the part you are looking at, or you will have dark edges.  Several fluorescent lights from the hardware store would be a good low budget way to test things.

Bruce

[BeardedBaron]

I'll look into those ideas.

Thank you!