I am trying to come up with a way to track points in 3D space.
What I have is two sets of 2D track points( X Y points ) that I
acquired from two camera views( one Vertical and the other Horizontal
). What I need is the corresponding 3D coordinates( X Y Z ) of the
object of interest. The Video images were taken from both the views
simultaneously.
I have scoured the net for papers and methodologies, but have not come
across similar work. I will put in points what I need :-
1. A method to do this conversion. 
2. Links to papers and websites referring to similar work.
3. I am using matlab for coding the algorithm, code for doing the
conversion would also help.
Thanks a lot in advance

---

Clarification of Question by ixerxesi-ga on 24 Jul 2005 06:42 PDT

I think it would be possible to make the points correspond in both the views.
And yes, the views are perpendicular for now. Did you come across
similar work anyone has done?

Presumably the two views you have are perpendicular to each other, so
that you can directly measure one of your co-ordinates in both
dimensions, and calibrate things that way? Your other two dimensions
may not have corresponding points to measure them from in both
pictures?

S

Not really sure what you are asking but:

Surely if you two views are perpendicular and correspond to the axes
(x,y) & (z,y), then you need only transpose one of the matricies and
ajoint the two such that you get a 3-d dimensional co-ordinate. i.e
the origin of camera one (0,0){x,y} should correspond with the origin
of camera two (0,0){z,y}

A much more complex problem is if you only have a single 2-d view with
corresponding (x,y) co-ordinates and no perception of depth. One way
of overcoming this problem is my refferencing an object that follows
brownian motion and using the known equations to calcuate its radius.
You could them compare the virtual radius and its brownian radius to
obtain a ratio. You could then use this ratio and combine it with the
known focal length of the camera to find it's 3rd dimension.