Multi-Camera 3D Tracking System


In this project, we aim to provide a strategy to build a Multi-Camera 3D Tracking System, being focused in the usage of Wiimotes as cameras. The first step in our strategy consists in making a deep study of the Wiimotes. We need to know how to connect them to the computer using a Bluetooth link and how we are able to retrieve information to them and use it in Matlab. By using concepts from epipolar geometry, we get information from 3D points from two different cameras and we are able to estimate the position and orientation of both cameras.

This project is only a first step of a Multi-Camera 3D Tracking System, since it can be adapted in order to support more Wiimotes, because for now it only supports two of them. Therefore, it can be used for any mobile objects on which IR LEDs can be attached to track them, making the system really versatile.


Real position of the WIimotes:

 Real Position of the Wiimotes

Reconstructed scenario by the algorithm:


With this example we can see that, after the step of calibration, the algorithm is capable to reconstruct the position of the cameras in the real world and reconstruct all the tracked points needed for the calibration (red circles).

Conclusion and Future Work

The proposed algorithm is only working for two Wiimotes. The next step is making the algorithm able to work with more Wiimotes. The calibration step could be also improved by adding a calibration pattern, in this way we could be able to know absolute position values. Finally, it would be interesting considering computing the error calibration, in order to know if the calibration performed is accurate enough.