Omnidirectional view and multi-modal streaming in 3D tele-immersion system

3D Tele-immersion (3DTI) technology allows full-body, multi-modal content delivery among geographically dispersed users. In 3DTI, user’s 3D model will be captured by multiple RGB-D (color plus depth) cameras surround- ing user’s body. In addition, various sensors (e.g., motion sensors, medical sensors, wearable gaming consoles, etc.) specified by the application will be included to deliver a multi-modal experience. In a traditional 2D live video streaming system, the interactivity of end users, choosing a specified viewpoint, has been crippled by the fact that they can only choose to see the physical scene captured by a physical camera, but not between two physical cameras. However, 3DTI system makes it possible rendering a 3D space where the viewers can view physical scene from arbitrary viewpoint. In this thesis, we present systematic solutions of omnidirectional view in 3D tele-immersion system in a real-time manner and in an on-demand streaming manner, called FreeViewer and OmniViewer, respectively. we provide a complete multi-modal 3D video streaming/rendering solution, which achieves the feature of omnidirectional view in monoscopic 3D systems