Author Topic: Can the WiiMote/ IR LED Array work through a 2-way mirror with naked fingers?  (Read 11653 times)

caeious

  • Guest
Greetings all,

I was wondering if anyone here knew if a WiiMote/IR LED array would be able to track naked fingers accurately with a 2-way mirror in between the WiiMote/IR LED Array and user's fingers?

For those that do not know exactly what a 2-way mirror is then read on. Essentially a 2-way mirror is an 1/8th to 1/4th inch thick glass or transparent acrylic plane in which both sides are treated with a 2-way mirror finish. The 2-way Mirror Finish essentially makes the glass, or acrylic, plane behave like a mirror when looking at it with in a brightly lit room. However when looking at the plane from a completely dark room, it behaves like a window.

To see what it looks like when taking a photograph through a 2-way mirror facing a brightly lit room look at the below picture



Here is the same picture taken with out the 2-way mirror inbetween camera and objects.



More info on 2-way mirrors can be found here: http://www.reflectionproducts.com/transparent-mirror.htm

I have to know if the IR setup (wiimote and IR LED array) will be able to track naked fingers (fingers with out reflective paper, etc) accurately at a distance of about 2 to 3.5 feet. The 2-way mirror website that I provided a link to above says that a TV remote will be able to turn on a TV through the 2-way mirror but we are talking about finger tracking here.

You thoughts?

-caeious




Offline cloudflint

  • *
  • Posts: 20
  • Karma: +4/-0
    • View Profile
The thing about a remote is that it is "one way", the signal goes from the remote to the tv. With finger tracking the infrared has to be able to get to the fingers and back again.

I think finger tracking with the wiimote without using markers is highly unlikley to be practical. As johnny lee states in his video "your fingers dont reflect much infra red light" so you would have to have a fairly massive infra red array. however this would mean that obviously the rest of your body would be reflecting infrared light as well. You could probably make an algorithm to work out which part was the fingers however since the blob detection for the wiimote is done with hardware i dont think you could do it with the wiimote.

regarding the one-way mirroe, if half the infared light is being reflected back I would have thought it would just be a case of increasing the power of the infrared light source so that enough light gets through to be picked up by the camera. However then the infrared light that is being reflected back may be picked up by the camera as well.

I suspect what may happen is that when you increase the power of the array loads of infrared will be reflected back by the mirror creating alot of "static" which would make it difficult to detect the markers. Though im not really sure.
« Last Edit: June 28, 2009, 12:51:57 PM by cloudflint »



caeious

  • Guest
Ok so how about using a powerful enough array that can track a user's fingers from around 2 to 3 feet away? I should mention I am not seeking perfect accuracy. That is to say, I am not seeking to just be able to track 1 finger alone. Tracking 3 fingers on one hand is acceptable and as we can see in Johnny Lee's finger tracking video....

[youtube]http://www.youtube.com/watch?v=0awjPUkBXOU[/youtube]

....at 2:12 Johnny says that there is a lot of extra noise (and we can only see that noise occurring from the left hand only. A computer program can be designed to filter out the bottom register points, or noise (the lower fingers of the left hand), and only use the top most register point (the pointer finger) to communicate with the rest of my art software (or any other software).

Now on the topic of whether or not Johnny Lee's IR LED array and Wii Mote IR camera will be able to send IR light out through the 2-way mirror and receive IR light through the 2 way mirror. Remember that if one is on the dark side of a 2-way mirror, the 2-way mirror will function as a window and not reflect light. I had mentioned in my original post that 2-way mirrors have been used to conceal TVs and that it is possible to turn the TV on with a standard remote control. So I guess that IR light can indeed be sent through a 2-way mirror if coming from the lit side. Since this is true, then I assume there would be no problem what so ever in sending and receiving IR light through both the darkened and lit sides of a 2-way mirror.

If you are not sure you understand anything I have just said then please let me know. I will explain it again with drawings. What are your thoughts? Do you think it is possible?

The thing about a remote is that it is "one way", the signal goes from the remote to the tv. With finger tracking the infrared has to be able to get to the fingers and back again.

I think finger tracking with the wiimote without using markers is highly unlikley to be practical. As johnny lee states in his video "your fingers dont reflect much infra red light" so you would have to have a fairly massive infra red array. however this would mean that obviously the rest of your body would be reflecting infrared light as well. You could probably make an algorithm to work out which part was the fingers however since the blob detection for the wiimote is done with hardware i dont think you could do it with the wiimote.

regarding the one-way mirroe, if half the infared light is being reflected back I would have thought it would just be a case of increasing the power of the infrared light source so that enough light gets through to be picked up by the camera. However then the infrared light that is being reflected back may be picked up by the camera as well.

I suspect what may happen is that when you increase the power of the array loads of infrared will be reflected back by the mirror creating alot of "static" which would make it difficult to detect the markers. Though im not really sure.



Offline cloudflint

  • *
  • Posts: 20
  • Karma: +4/-0
    • View Profile
Quote
....at 2:12 Johnny says that there is a lot of extra noise (and we can only see that noise occurring from the left hand only. A computer program can be designed to filter out the bottom register points, or noise (the lower fingers of the left hand), and only use the top most register point (the pointer finger) to communicate with the rest of my art software (or any other software).

The thing is the wiimote is only able to track 4 points, this is built in to the wiimotes hardware and as far as i know there is no way of changing it. As far as i know there is no way of getting the raw input from the wiimote camera, thus it is impossible to alter the algorithm to work with high levels of static.

Quote
Now on the topic of whether or not Johnny Lee's IR LED array and Wii Mote IR camera will be able to send IR light out through the 2-way mirror and receive IR light through the 2 way mirror. Remember that if one is on the dark side of a 2-way mirror, the 2-way mirror will function as a window and not reflect light. I had mentioned in my original post that 2-way mirrors have been used to conceal TVs and that it is possible to turn the TV on with a standard remote control. So I guess that IR light can indeed be sent through a 2-way mirror if coming from the lit side. Since this is true, then I assume there would be no problem what so ever in sending and receiving IR light through both the darkened and lit sides of a 2-way mirror.

The thing is, two way mirrors work by reflecting half the light that is cast on them back. So looking from a bright room to a dark room, the light that is being reflected back completely obscures the small amount of light that is coming from the dark room. And on the dark room side, the half of the light from the bright room that is getting through completely obscures the small amount of light that is being reflected back.

So if you shine a infrared light through a two way mirror, regardless of what side you are on half of the infra red light will go through and half will be reflected back. This means that you could operate a TV from one side with a remote on the other as some of the signal from the remote will go through. With the array and camera on the same side of the glass however, half of the light from the array will be reflected straight back into the camera, effectively drowning out any light that may be reflected back from objects on the other side of the glass. at least that's my theory.

My suggestions would be to maybe use a tracking method which does not require a specific light source. the reason using infra red light for the tracking works so well is that there isn't large levels of it in our environment, so we know that anything that is reflected back will be from our markers and thus do not require complex algorithms. since you would have to have fairly complex algorithms anyway perhaps you should look into having a web cam on the dark side of the glass and use skin tone detection?




caeious

  • Guest
Quote
The thing is the wiimote is only able to track 4 points, this is built in to the wiimotes hardware and as far as i know there is no way of changing it. As far as i know there is no way of getting the raw input from the wiimote camera, thus it is impossible to alter the algorithm to work with high levels of static.

But on Johnny's video I see more than 4 points on the screen. Or is that because he is moving his fingers so fast? Well even if it can track only 4 points that is ok. A software filter can be set up to eliminate the far left side (or bottom) register points for the left hand and the far right (or bottom) register points for the right hand.

Quote
The thing is, two way mirrors work by reflecting half the light that is cast on them back. So looking from a bright room to a dark room, the light that is being reflected back completely obscures the small amount of light that is coming from the dark room. And on the dark room side, the half of the light from the bright room that is getting through completely obscures the small amount of light that is being reflected back.

I do not really think this is true. How can half of the light be getting through to the dark room side? Based on the photo in my original post, it sure looks like most of the light is getting through to the dark room side where the IR setup will be located. Oh I see what you mean, you are saying that the IR light bouncing off the fingers will just be reflected back off the light room side of the 2 way mirror. Hmm.

And then you say:

Quote
With the array and camera on the same side of the glass however, half of the light from the array will be reflected straight back into the camera, effectively drowning out any light that may be reflected back from objects on the other side of the glass. at least that's my theory.

Now I am wondering if IR light goes through completely or gets reflected. Based on the fact that a TV remote's IR light signal can go through, I assume that IR light will pass through a 2-way mirror.

Is there anyone here that can run a test to see if what I am proposing will work? I would need this test to be filmed and uploaded to Youtube. I would pay for a 2 way mirror and costs of taping the process. Cloudflint, if you are interested please let me know.

-caeious




Offline cloudflint

  • *
  • Posts: 20
  • Karma: +4/-0
    • View Profile
Hmm i ccould probably test it if i can get ahold of a two way mirror. I shall look into it.



Offline wizzed

  • *
  • Posts: 5
  • Karma: +0/-0
    • View Profile
But on Johnny's video I see more than 4 points on the screen.
As far as i understand it from reading around, the wiimote can only detect 4 blobs at a time (i.e. frame).
If, however, the intensities are not sufficiently distinct you have no guarantee that you get the same 4 blobs in two consecutive frame.

HTH,
wizzed