3D Video (3DV) and free viewpoint video (FVV) demand some form of 3D scene representation been encoded efficiently. 3D scene representation determines coding, compression and transmission. In general, two methods for 3D scene representation can be found in the literature: Image based modeling and geometry based modeling. Image based modeling does not use 3D Geometry at all. If image based modeling is used for FVV, Virtual intermediate views are generated from natural camera views by interpolation. Geometry based methods allow full freedom of virtual view rendering but demand an accurate 3D reconstruction of the scene. As opposed to typical animation sequences which consist of dynamic meshes, reconstructed dynamic 3D scenes, which are used in geometry based modeling, typically consist of time varying meshes (TVMs). Dynamic meshes are meshes that have constant vertex count and constant connectivity across frames but variable geometry (vertex coordinates). On the other hand, time varying meshes are meshes that their vertex count and connectivity change over time so that geometry, connectivity and vertex count are variable across frames. ITI has already extensively worked on real time 3D scene reconstruction and is currently working on real time coding & compression of geometry based dynamic 3D scene representation with applications in data transmission for live 3d immersive environments.
3D Living Interactions through Visual Environments