Performance Evaluation and Optimization of the Teleimmersion System
Gregorij Kurillo and Ruzena Bajcsy
National Science Foundation
Teleimmersion combines virtual reality and teleconferencing to allow people from around the world to collaborate in real time inside a shared virtual environment. The teleimmersion system at the CITRIS lab consists of 48 cameras in 12 stereo clusters. Images are processed by 12 computers running simultaneously and sending data via Internet II connection to near and remote rendering computers.
This work aims to evaluate different aspects of this multi-camera vision system and address the problems which are currently reducing its performance. The teleimmersion system will be analyzed top-down by assessing noise of camera CCD , comparing the performance of individual cameras, analyzing calibration accuracy , calculating volume coverage of the stereo clusters , and evaluating the performance of the stereo reconstruction. A toolbox for simulation of multiple camera coverage will be developed to assist with the optimal positioning of the camera clusters. From the results of the evaluation, improvements will be implemented to enhance the performance of the teleimmersion system.
Figure 1: Volume coverage analysis of a stereo cluster (3D view)
Figure 2: Volume coverage analysis of a stereo cluster (top view)
- G. Kamberova and R. Bajcsy, "Sensor Errors and the Uncertainties in Stereo Reconstruction," K. W. Bowyer and P. J. Phillips, eds., Empirical Evaluation Techniques in Computer Vision, IEEE Computer Society Press, 1998, pp. 96-116.
- W. Zhao and N. Nandhakumar, "Effects of Camera Alignment Errors on Stereoscopic Depth Estimates," Pattern Recognition, Vol. 29, 1996, pp. 2115-2126.
- S. Drouin, R. Poulin, P. Hébert, and M. Parizeau, "Monitoring Human Activities: Flexible Calibration of a Wide Area System of Synchronized Cameras," Proc. Int. Conf. Vision Interface, June 11-13 2003, pp. 49-56,