0
$\begingroup$

I hope this the right place for this question, not sure. I have a Asus Xtion Pro live camera, so an IR structured light sensor, on a shop window. When the shop window glass mirrors (there is more light inside than outside the shop), I see this mirrored image also in the depth map. It's clearly an artifact, but how is it possible?

enter image description here

enter image description here

Ignore the holes in the depthmap (black areas) for the moment, it's the not the focus of this question, but look at the red circle, this is a table, that is located inside the store (as you can see in the rgb frame), it should not be present in the depthmap. As far as I know a structured light sensor project a light pattern on the scene and through a camera read how the pattern is deformed on the different surfaces (near, far) to calculate the depth (distance of every point in the detphmap). Of course the table doesn't exist in the scene, so should not exist in the depthmap too! Any idea?

$\endgroup$
  • $\begingroup$ Can I please ask if this was resolved? $\endgroup$ – A_A Feb 27 '18 at 10:59
  • $\begingroup$ Not really, but helped in understanding what the problem is.. $\endgroup$ – rok Feb 27 '18 at 13:06
2
$\begingroup$

Structured or not, light gets reflected and the sensor has no way of telling if the scene is imaged directly or by reflection.

To try and reduce the effect of reflections from the glass, you might want to use a polariser. Photographic polarising filters have lots of useful applications and one of them is reducing glare when you are trying to shoot artifacts within a museum glass case (for example).

You can find really cheap polarising filters in the second-hand boxes in camera shops, or buy polarising sheets which you can cut into various shapes.

One thing that would be good to note however is that the camera may already be using a polarising element internally which might interfere with an additional polariser which will worsen the situation.

Hope this helps.

$\endgroup$
  • $\begingroup$ tried with a polarising sheet but this worsen the situation, probably the camera has some internal polarizing element as you suggested. Anyway this is a very borderline situation, and I think the only solution is to put the camera directly on the glass, to avoid the mirroring effect.. $\endgroup$ – rok Feb 27 '18 at 13:09
  • $\begingroup$ @rok Thanks for letting me know. The polariser reduces light intensity somewhat too, so then, it might be getting to the point where certain regions are dropped altogether because the sensor does not have enough light to make a decision within the specs of the device. All the best $\endgroup$ – A_A Feb 27 '18 at 14:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.