3d object recognition

This process is called photogrammetry... If you have photos taken from very exact locations you know... you simply use a reverse camera transform... to make screen points into lines from your photos, into 3D space... if the photos are taken from specific locations, known to you... you can intersect these lines into 3D points. They can then be used to form a model... without exact placement.. the whole thing becomes MUCH more complicated... rather than just plain onld 3D geometry you have to delve into the murky waters of statistics!

Thanks a million for the term! I can now go and do some lookups. Really appreciate the response.

I might be an amateur programmer, but i'm a proffessional 3D artist.

What you aiming for is 3D projection.
For this you atleast need 3 cameras on a x,y,z setup around the object.

Now you need to code your programm in a way that it can pick recognisable points in an image.

You say you want your remote cntrol car in 3D with depth etc etc on your screen.

You programm needs to position a point in 3D space viewed from all 3 viewpoints/cameras.

f.i. Your programm picks a point in your "remote control car image"
It picks the same point viewed through the x axis, aswell as the y axis and the z axis.
Now say your point in 3D space lies on the left headlight of your car.
You need to get that point on the headlight locked in all 3 viewports so you can "track" it.

To make it easier for you, you could put little "red X stickers" on your car for the 3 cameras to be able to see those stickers on your car, so that your programm can pick those stickers as 3D points.

So now you have many points in space with x,y,z coordinates.
This shhould give you a startoff to make it 3D or give it some depth of perception.

Good luck.

[edited by - TheRademaker on August 12, 2002 3:01:23 AM]