5

I am trying to build a protective enclosure for Microsoft's Kinect v2, which emits (and perceives) IR in the 827–850 nm range. In order to fully enclose the device without blocking those signals, I need some transparent and hard material that will let infrared pass through.

I have tried acrylic, glass and crystal from the local glass cutter, but I can't get good results from any of those materials.

The best result I got was when I stuck the pane flush to the Kinect's front, otherwise the IR camera just can't see anything. However, technically I can't do that because I need to allow the Kinect to tilt, and I can't have the pane always flush to the Kinect.

The enclosure will be completely opaque apart from the front pane (which is what I'm looking the material for).

Air
  • 3,221
  • 4
  • 24
  • 46
Kal
  • 151
  • 4
  • Hi Kal, welcome to engineering.SE. Can you please clarify how the Kinect will will be mounted relative to the transparent material? You seem to indicate that it will be in a box; will all of the walls be made of the same transparent material? – Chris Mueller Apr 11 '16 at 12:00
  • @ChrisMueller Thanks for the reminder. Clarification added at the end of the question. – Kal Apr 11 '16 at 15:07
  • Does the front panel need to be transparent in the visible spectrum, or only to IR? – Air Apr 11 '16 at 16:43

2 Answers2

5

Acrylic and glass do let 800-900 nm IR radiation through quite well (see, e.g., here).

However, your problem is probably that part of the light gets reflected at each of the surfaces. This is unavoidable, though it can be reduced by some anti-reflective coatings, but those are not available off-the-shelf as they need to be designed according to material and wavelength.

Further problems will be caused by the refraction, which will distort the path of the light. This may interfere with the calibration and focusing of the Kinect's optical parts.

Air
  • 3,221
  • 4
  • 24
  • 46
jpa
  • 1,845
  • 7
  • 12
  • I suspect that you are correct about surface reflections causing an issue. Refraction, on the other hand, should not cause a problem unless some of the surfaces are curved. – Chris Mueller Apr 11 '16 at 11:58
  • @ChrisMueller If I've understood correctly, Kinect uses an IR projector that projects structured light pattern on the scene, which is then detected by the IR camera. Inserting a transparent plate in between will change the relative position of the projected pattern. I do not know whether Kinect's algorithms are smart enough to compensate for this or not. – jpa Apr 11 '16 at 12:27
  • @ChrisMueller actually, refraction can cause a angle-dependent distortion even for flate plates, because the ray paths for objects well off the optic axis enter at different angles and have different effectivd optical path lengths thru the plate. But I agree that this sure looks like a reflection problem. – Carl Witthoft Apr 11 '16 at 13:23
  • @CarlWitthoft In theory, yes, refraction will affect the optical path. But in practice, it doesn't seem to be noticeable, and probably not to an extent that affects my usage. – Kal Apr 11 '16 at 15:10
1

As jpa points out, the issue is not with the transparency of the materials that you are considering. I suspect that the issue is most likely caused by light being reflected directly from the IR emitter into the camera.

The intensity of light specularly reflected from a surface very close to the camera will be orders of magnitude brighter than the faint amount of light diffusely reflected from clothing and other objects which are much farther away. This will saturate the image sensor in the IR camera and make it impossible to see the IR pattern from other objects. Imagine trying to see the light of a small LED through a 100 W light bulb.

If I am correct, then the solution is fairly easy. You simply need to tilt the transparent material at a fairly steep angle so that the reflected light is sent to a part of the enclosure where you can put a material which strongly absorbs IR. Exactly how steep the angle needs to be depends on the details of the emitted IR pattern and the position of the screen, but I suspect that it will need to be at least 25°. If you tilt the material all the way to ~55° then you may be able to take advantage of Brewster's angle to significantly reduce the amount of reflected light, but the effectiveness depends on the polarization state of the emitted light and the direction of the tilt.

Chris Mueller
  • 6,386
  • 1
  • 26
  • 54
  • I'm also starting to suspect that's the cause of my problems. I think there may be another solution, which is to partition the space between the pane and the Kinect, using some opaque material, so to prevent the IR reflection from reaching the sensor. – Kal Apr 12 '16 at 01:55