Hi,

I'm just starting a dissertation in mixed reality gaming, and I'm
investigating ARToolkit and the possibilities it presents regarding
tracking game players. I was thinking of using a combination of ARToolkit
and GPS, with ARToolkit resolving the position of the player to a good
degree of accuracy.

What I'm asking is this: am I correct in thinking that it's possible to
use the libraries to determine the camera's position given that you know
the exact positions of the markers? I read some docs, but they appear to
be half done and out of date, so I was wondering if I'm on the right track
and whether what I'm asking the system to do is feasible with ARToolkit.

Am I also correct in thinking that arDetectMarker() and arGetTransMat() 
will enable me to extract the positional data of the camera, given that I
know the locations of the markers in the physical world?

Thanks.

later,
Stuart
---
email: stuart@t ............

my pointless, self-congratulatory vanity site is located here:
http://www.nontrivial.uklinux.net

Hi,

>What I'm asking is this: am I correct in thinking that it's possible to
>use the libraries to determine the camera's position given that you know
>the exact positions of the markers?

Actually, ARToolKit calculates marker position in camera coordinates
by 3x4 transformation matrix. In order to get cmera position in marker
coordinates, you have to invert the matrix. This inverted matrix represents
transformation from camera to marker. therefore translation component of the
matrix means camera position in marker coodinates.

>Am I also correct in thinking that arDetectMarker() and arGetTransMat() 
>will enable me to extract the positional data of the camera, given that I
>know the locations of the markers in the physical world?

You have to take care of the accuracy.
Basically ARToolKit is designed for image overlay.
It is not designed for 3D measurement system.
Anyway you have to calibrate your camera very carefully.

-- 
------------------------------------------------------------------
  Hirokazu Kato
  Faculty of Information Sciences
  Hiroshima City University               Phone: +81-82-830-1705
  Email: kato@s ........................     Fax: +81-82-830-1435
  URL: http://www.sys.im.hiroshima-cu.ac.jp/people/kato/

Hi Kasim,
thank you for the reply! I'm a new subscriber to this list so I don't have
access to previous posts.
It might be very simple, but how does the 3x4 matrix describe the camera
transformation? I'm certainly not very into matrices... Is it some kind of a
origin and tree vectors?

Regards, Teis


> Hi Teis,
> This might help you: A reply of Kato to a previous query.
>
> Kasim
>
> ----- Original Message -----
> From: "Hirokazu Kato" <kato@s ........................>
> To: "Stuart" <stuart@t ............>
> Cc: <artoolkit@h ..................>
> Sent: Thursday, October 03, 2002 5:17 AM
> Subject: Re: Position tracking
>
> > Hi,
> >
> > >What I'm asking is this: am I correct in thinking that it's possible to
> > >use the libraries to determine the camera's position given that you
know
> > >the exact positions of the markers?
> >
> > Actually, ARToolKit calculates marker position in camera coordinates
> > by 3x4 transformation matrix. In order to get cmera position in marker
> > coordinates, you have to invert the matrix. This inverted matrix
> represents
> > transformation from camera to marker. therefore translation component of
> the
> > matrix means camera position in marker coodinates.
> >
> > >Am I also correct in thinking that arDetectMarker() and arGetTransMat()
> > >will enable me to extract the positional data of the camera, given that
I
> > >know the locations of the markers in the physical world?
> >
> > You have to take care of the accuracy.
> > Basically ARToolKit is designed for image overlay.
> > It is not designed for 3D measurement system.
> > Anyway you have to calibrate your camera very carefully.


 

I think a book on computer graphics will help you in that case. All books
have a chapter on transformations. The fourth column of the matrix can be
seen as a translation vector, the rest as a 3x3 rotation matrix.

Some references:
http://www.sys.im.hiroshima-cu.ac.jp/people/kato/etc/ATK-coordinates.jpg
http://dlib.computer.org/conferen/iwar/0359/pdf/03590085.pdf (you will need
a IEEE Computer Society password, or a site subscription to the IEEE digital
library).

Kasim

----- Original Message -----
From: "Teis Draiby" <teis@d .........>
To: <artoolkit@h ..................>
Sent: Sunday, December 15, 2002 2:54 AM
Subject: Re: Position tracking


> Hi Kasim,
> thank you for the reply! I'm a new subscriber to this list so I don't have
> access to previous posts.
> It might be very simple, but how does the 3x4 matrix describe the camera
> transformation? I'm certainly not very into matrices... Is it some kind of
a
> origin and tree vectors?
>
> Regards, Teis
>
>
> > Hi Teis,
> > This might help you: A reply of Kato to a previous query.
> >
> > Kasim
> >
> > ----- Original Message -----
> > From: "Hirokazu Kato" <kato@s ........................>
> > To: "Stuart" <stuart@t ............>
> > Cc: <artoolkit@h ..................>
> > Sent: Thursday, October 03, 2002 5:17 AM
> > Subject: Re: Position tracking
> >
> > > Hi,
> > >
> > > >What I'm asking is this: am I correct in thinking that it's possible
to
> > > >use the libraries to determine the camera's position given that you
> know
> > > >the exact positions of the markers?
> > >
> > > Actually, ARToolKit calculates marker position in camera coordinates
> > > by 3x4 transformation matrix. In order to get cmera position in marker
> > > coordinates, you have to invert the matrix. This inverted matrix
> > represents
> > > transformation from camera to marker. therefore translation component
of
> > the
> > > matrix means camera position in marker coodinates.
> > >
> > > >Am I also correct in thinking that arDetectMarker() and
arGetTransMat()
> > > >will enable me to extract the positional data of the camera, given
that
> I
> > > >know the locations of the markers in the physical world?
> > >
> > > You have to take care of the accuracy.
> > > Basically ARToolKit is designed for image overlay.
> > > It is not designed for 3D measurement system.
> > > Anyway you have to calibrate your camera very carefully.
>
>

 

hi teis

or have a look at the opengl redbook.

http://fly.cc.fer.hr/~unreal/theredbook/appendixg.html

christoph

Teis Draiby wrote:
> Hi Kasim,
> thank you for the reply! I'm a new subscriber to this list so I don't have
> access to previous posts.
> It might be very simple, but how does the 3x4 matrix describe the camera
> transformation? I'm certainly not very into matrices... Is it some kind of a
> origin and tree vectors?
> 
> Regards, Teis
> 
> 
> 
>>Hi Teis,
>>This might help you: A reply of Kato to a previous query.
>>
>>Kasim
>>
>>----- Original Message -----
>>From: "Hirokazu Kato" <kato@s ........................>
>>To: "Stuart" <stuart@t ............>
>>Cc: <artoolkit@h ..................>
>>Sent: Thursday, October 03, 2002 5:17 AM
>>Subject: Re: Position tracking
>>
>>
>>>Hi,
>>>
>>>
>>>>What I'm asking is this: am I correct in thinking that it's possible to
>>>>use the libraries to determine the camera's position given that you
> 
> know
> 
>>>>the exact positions of the markers?
>>>
>>>Actually, ARToolKit calculates marker position in camera coordinates
>>>by 3x4 transformation matrix. In order to get cmera position in marker
>>>coordinates, you have to invert the matrix. This inverted matrix
>>
>>represents
>>
>>>transformation from camera to marker. therefore translation component of
>>
>>the
>>
>>>matrix means camera position in marker coodinates.
>>>
>>>
>>>>Am I also correct in thinking that arDetectMarker() and arGetTransMat()
>>>>will enable me to extract the positional data of the camera, given that
> 
> I
> 
>>>>know the locations of the markers in the physical world?
>>>
>>>You have to take care of the accuracy.
>>>Basically ARToolKit is designed for image overlay.
>>>It is not designed for 3D measurement system.
>>>Anyway you have to calibrate your camera very carefully.
> 
> 


-- 
Christoph Ganser
ETH World                      mailto:ganser@e ...............
HPI F23                           http://www.ethworld.ethz.ch/
CH-8093 Zuerich                          phone:+41 1 633 2902

> or have a look at the opengl redbook.
> 
> http://fly.cc.fer.hr/~unreal/theredbook/appendixg.html

This is all very well, but are you saying that if you do the inverse
calculation of the matrix M on the top left 3x3 part of the config->trans
matrix, the camera rotations can be discovered? I'm unsure as to the inner
workings of ARToolkit and the way the rotation matrix is set up.

I'm currently trying to extract the camera rotations for the x, y and z
plane (i.e. i would like to recover three different rotation angles). What's 
the best way to get these individual angles?


later,
Stuart
---
email: stuart@t ............

my pointless, self-congratulatory vanity site is located here:
http://www.nontrivial.uklinux.net

Thanks for the links, Kasim and christoph. Just what I needed.

Teis