The ScanTrack Virtual Environment

摘要

We are developing the Scan&Track virtual environment (VE) using multiple cameras. The Scan&Track VE is based upon a new method for 3D position estimation called the active-space indexing method. In addition, we have also developed the geometric-imprints algorithm for significant points extraction from multiple camera-images. Together, the geometric-imprints algorithm and the active-space indexing method, provide a promising and elegant solution to several inherent challenges facing camera-based virtual environments. Some of these are: (a) correspondence problem across multiple camera images, (b) discriminating multiple participants in virtual environments, (c) avatars (synthetic actors) representing participants, (d) occlusion. We also address a fundamental issue: can virtual environments be powerful enough to understand human-participants.