Author Topic: Isn't it better to use the reflectors method?  (Read 14409 times)

Offline mknewman

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Reply #15 on: February 01, 2008, 01:39:50 PM
How about corner reflectors like we left on the moon!



Offline cpetersen

  • *
  • Posts: 10
  • Karma: +0/-0
    • View Profile
Reply #16 on: February 02, 2008, 10:54:24 AM
Interesting idea. As I've posted elsewhere I'm trying to come up with a head tracking system for my quadriplegic sister. After building a IR bar and getting a BT stack for the windows side of my iMac and doing some testing I've come to the conclusion that head tracking is not really what I want to achieve. What I need is face tracking. My aim is to track where a face is pointing (with head stationary) and move a mouse on the screen. Back on topic here, I'll have to try this idea and see if I can get the resolution I need to do the face tracking. I've started work on the software and see if I can null out the head position and register the IR point positions in space.

Thanks
Carl



Offline troy59

  • *
  • Posts: 123
  • Karma: +2/-0
    • View Profile
Reply #17 on: February 03, 2008, 07:49:45 PM
ok, what do you mean by "face tracking" You say that instead of moving your head it is actually stationary. That's kinda confusing, because the only way you can move a mouse with your head is by moving it. If you can explain, that'd be great.



Offline cpetersen

  • *
  • Posts: 10
  • Karma: +0/-0
    • View Profile
Reply #18 on: February 03, 2008, 08:04:30 PM
ok, what do you mean by "face tracking" You say that instead of moving your head it is actually stationary. That's kinda confusing, because the only way you can move a mouse with your head is by moving it. If you can explain, that'd be great.
If you hold your head still and rotate it left, right, or nod your head you haven't actually moved your head in space only its orientation. You could argue the semantics but the current demos i've been running do a great job of tracking your head in 3D space but are relatively poor at tracking where your nose is pointed.

I need that for my application and I think it's just a matter of SMOP (small matter of programming). Certainly doable. I'm having more success with small reflectors rather than the IR glasses approach.

HTHs






Offline troy59

  • *
  • Posts: 123
  • Karma: +2/-0
    • View Profile
Reply #19 on: February 06, 2008, 10:49:09 AM
oh, i understand, thats what i was thinking of in your origional post, it would be to tiring having to move your head like that.



Offline Timo

  • *
  • Posts: 3
  • Karma: +0/-0
    • View Profile
Reply #20 on: February 06, 2008, 01:18:40 PM
I personally prefer the reflector method.
You have the following advantages:
- need no battery at tracked device (e.g. head)
- using bigger reflectors you get subpixel resolution (I don't know if this works with the wiimote)

I use reflector tape that you can get in stores (probably from scotch) - in Germany in buy it in "Obi Baumarkt". Then I put this on small wood balls and finished are my markers.
Then you need IR-light directly from the camera - the wiimote. It will not work, if you put the light somewhere else! Depending on the distance you work with you need the right amount of ir-leds. In electronic stores you might find prebuilt ir-lights, usually for night vision security application. But you can also build it yourself, but buy leds with a wide opening angle as most ir-lights are spotlights. In Germany you can get a nice little ring of ir-leds from "Conrad".

Using your eyes will not work, as they do not reflect enough. It is only used in very special eye-tracking systmes where the camera more or less only sees the eyes.

Cheers,

Timo



Offline EuroMateo

  • *
  • Posts: 16
  • Karma: +1/-0
  • Please join my project CHCIMC
    • View Profile
Reply #21 on: February 15, 2008, 05:16:06 AM
Quote: "I already posted this on the project blog before I found this site, but the benefit to using reflectors for head tracking, as I imagine the OP envisioned, is the fact that reflectors don't require a power source.  Taking it a step farther, one could simply coat each lens on a pair of glasses (better yet, stereoscopic glasses) with a substance which is transparent in the visible spectrum and reflective in the IR spectrum."

.. F**K!
ur right. GOD DAMN IT! Is there anyway around this? anyone?!?!

and also in re; to "eyes not reflecting enough light"

Well then just change the sensitivity.. or increase the light output. :)

Btw. If you have programing knowledge.. or can in anyway modify Lee's original program to recognize eyeballs only and automatically ignore all other reflections, I WILL PAY YOU! I am very desperate. Please help! Email; [email protected]


Offline EuroMateo

  • *
  • Posts: 16
  • Karma: +1/-0
  • Please join my project CHCIMC
    • View Profile
Reply #22 on: February 15, 2008, 05:30:07 AM
Solution; A combination of wiimotes to allow more than just 4 points. Im sure someone in these forums has done this already.