Jump to content
  • Advertisement
Sign in to follow this  
abolfoooud

Back projecting on-screen point for Augmented reality use

This topic is 4231 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

Hi, I am trying to do an augmented reality application using special markers i designed. I managed to get the corners and the center of the marker on screen (2D). now i need to do the real work: augmenting 3d models on the marker. I want to understand how this works? i want to undestand how to back project these 2D points into their real location in the 3D world. i am not using ARToolkit. i need to create the function that does this myself. can any one give me an insight of how to accomplish this task? any reference, explanation, books, articles...anything will be appriciated thanx in advance AF

Share this post


Link to post
Share on other sites
Advertisement
well, on the AR Toolkit website there's a collection of papers related to the algorithms involved. Searching google and especially citeseer might reveal tons of papers on marker-based tracking, shape from motion etc.

I havent done any work in AR but in a domain somewhat related to it.

First determine the camera's intrinsic and extrinsic parameters (e.g. tsai for coplanar surfaces)
-> camera calibration.

With the calibration data in the pocket it's possible to project the 2d image plane positions back to world-space (e.g. on a plane).

visionbib.com contains alot of information on that.





Share this post


Link to post
Share on other sites
You lose information when you transform from 3D to 2D. There are a couple ways to reconstitute that data.

1) there's an openGL specific way to reverse project; search around on the forums. i don't know that method

2) you cast a ray through your world using the screen point transformed to the near and far clip planes as endpoints. Collide that ray with your level geometry to the appropriate resolution. If the ray doesn't collide with anything then the point is the far-clip plane endpoint.

Transforming that point to the near and far clip plane is trivial since you just take the relative screen offset and use that on the planes (whose bounds you can extract from the camera transform and the projection matrix). You can find details on this in the forums or google: I don't have the links handy.

-me

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!