Author Topic: 9-point calibration  (Read 11152 times)

Offline Hoaxjak

  • *
  • Posts: 16
  • Karma: +0/-0
    • View Profile
on: February 09, 2009, 01:11:33 AM
just saw the demo video for that "competing" interactive whiteboard system in the other thread, and while it didn't look appealing at all ($950? what!?), one thing stood out......

the 9 point calibration. Is there any way to retrofit Smoothboard with a 9 point calibration, or at least something more than 4 point? 9 point, i think, will greatly improve tracking resolution and fine-detail accuracy.



Offline Wiweeyum

  • *
  • Posts: 203
  • Karma: +7/-0
    • View Profile
Reply #1 on: February 09, 2009, 02:34:33 AM
I don't know... four point seems plenty good enough for me with a rectangular screen. Nine point would be good if you were on something like a fish-eyed screen, but if it's a quadrilateral, four is all you really need for an accurate calibration.

That said, it does look impressive and just says "I'm more professional than you".


~"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction."~
- EF Schumacher


Offline benpaddlejones

  • Support Admin
  • *
  • Posts: 1178
  • Karma: +49/-0
    • View Profile
    • Inter.Connect.Ed
Reply #2 on: February 09, 2009, 04:35:30 AM
Hoaxjak
Your actually increasing the opportunity for user error.

If you are calibrating two flat 2D surfaces 4 points is as accurate as 9 points. 9 points is more likely to be affected by user input error, skewing all calibration points.

If either of the surfaces are not a flat 2D surface then yes more calibration points will improve accuracy.

The 9 point is used by smartboard because of the natural variations in the matrix that seperates the two layers. I would suggest 9 points is used in the other system becuase the lense is cheap, changing the IR-camera's view or more likely they copied smartboard and don't really understand the mathematics.

What matters in the 4 point calibraton is taking a few extra microseconds to make sure your IR-Pen is in teh centre of the cross hair.

benpaddlejones :)


See my full profile and links on my Google Profile: http://www.google.com/profiles/benpaddlejones


Offline benpaddlejones

  • Support Admin
  • *
  • Posts: 1178
  • Karma: +49/-0
    • View Profile
    • Inter.Connect.Ed
Reply #3 on: February 10, 2009, 05:37:26 AM
Hoaxjak/Wiweeyum
I was playing with ebeams with the Oz distributor today, I asked why they used a 9 point calibration. There helpful sales team had no idea, but in a later conversation unwittingly told me.

I add that I always thought the ebeam provided less functionality than other commercial IWB's. Infact for the price point (about $1000 for the mobile version) it actually performes above expectations. Although it is still 20x as expensive as the Wiimote IWB so we are still smiling smugly.

I assumed the ebeam worked on a IR-detection system similar to the Wiimote. infact it doesn't:

When an eBeam pen first touches the board it emits a flash of infra-red which is picked up on a sensor in the centre of the eBeam receiver. The pen when being used then emits a series of ultrasonic signals which can be heard as a slight ‘buzzing sound’; this signal is received on the poles of the receiver. The eBeam unit then calculates the difference in time between the two signals and triangulates the position of the stylus or marker sleeve on the pre-calibrated area. The principle is not too dissimilar to that of a child counting the difference in time between lightening and thunder to locate the centre of a storm.

So the nine calibrations are required to provide an accurate calibration as only four points would not provide enough data for triangulation of the ultrasonic signal. To take the locating thunder analogy: if lightening strikes and anyone in the area pointed to where they thought the lightening touched down, the sum of 9 people distributed across a large area will provide a more accurate location than just 4.

Because the Wiimote is a 2D flat surface that detects a IR-Source and that location is calibarted to another flat 2D surface 4 points is all that is needed. Even in keystone displays the algorythm stands accurate.

As I suggested in my earlier post, 9 points will actually diminish the quality of the calibration for the Wiimote.

benpaddlejones :)

« Last Edit: February 10, 2009, 05:41:44 AM by benpaddlejones »


See my full profile and links on my Google Profile: http://www.google.com/profiles/benpaddlejones


Offline Wiweeyum

  • *
  • Posts: 203
  • Karma: +7/-0
    • View Profile
Reply #4 on: February 11, 2009, 04:00:29 AM
Ha! Beautifully put. That's great info to know. Would excess noise screw up the accuracy then? Heh, and I mean actual noise being the extra noise, just as the sun creates IR noise for the wiimote.


~"Any intelligent fool can make things bigger, more complex, and more violent. It takes a touch of genius -- and a lot of courage -- to move in the opposite direction."~
- EF Schumacher


Offline bubka

  • *
  • Posts: 26
  • Karma: +2/-0
    • View Profile
Reply #5 on: February 16, 2009, 05:31:33 PM
I have personally used it, it is pretty spot on.


[youtube]http://www.youtube.com/watch?v=B4ccysJXhZ4[/youtube]


I guess if one is off then the overall calibration is not effect as much.  I know if I got one of the 4 wrong, I am in trouble on that corner of the board.

Too bad I will not be getting this in the old free version of smoothboard :(