Calibration
We developped different simples and intuitives method for calibration of elements on our environment (publish in [VRIC2001]). With an user-bases approach, system ask user to realized some manipulations in few steps. We name our method 'plug'n calib and play' in the spirit of calibration of joystick on PC.
We obtain a maximum error in order of 1 cm, that is sufficient for non-critical applications. The stability and accuracy of Optotrack give large better result that the old system, based on magnetic tracker.
Results
We show here some result of our calibration method. This image provide of a video see-through system.
Internal parameter of camera are estimated with vision method (based on calibration method of ARToolkit). External parameter are calculated on real time, based on position of user head. A Static calibration phase in necessay for compute transformation between optical tracker position (CS defined by 3 leds place on HMD) and camera position. They are estimated with our methods (derived method of [Bajura]).
References
Merging Real and Virtual Environments with Video See-Through Head-Mounted Displays. 1997 PhD Thesis. Michael A. Bajura.
Visual Integration : Realistic Rendering
The two main points are occultations between virtual and real object, and the shadow of virtual/real on virtual/real objects.
Occultations
For the occultation, with specific method, we obtain a geometry representation of real object, and is position on the table, and we after just do a z-rendering of the object.
References
|