• telepresence
• multi camera
• depth-map compression
• augmented reality
• AR
• 3D capture
• virtual reality
• 3D
• VR

This thesis investigates some of the key aspects of a 3D telepresence system. It starts from the capturing methods analyzing different technologies and discussing their characteristics. It deals with the capturing devices, their calibration, the capturing software and the software quality-enhancers. These aspects linked together enable the system to capture a real-time 3D scene.
Then the thesis inspects the rendering techniques of the acquired 3D scene. This aspect is analyzed independently from the chosen rendering device. Thus, it can be applied for example to an HMD, a see-through device, a stereoscopic monitor or a standard rendering device.
Moreover, scene compression is analyzed to enable a telepresence system to be adopted on a network with a limited bandwidth such as internet.
Instead of finding new compression algorithms, this thesis tries to exploit the widely adopted standard video compressors to compress both video streams and geometry information. This approach aims to exploit the great work done for video compression in the past years.
Finally, it proposes a system architecture for telepresence systems. It aims to enable the development of telepresence systems that exploit hardware acceleration support. There are already a lot of cheap devices with such hardware support, ranging from laptop to single-board computers. The approach of this thesis is to make them exploiting their hardware to execute a telepresence system.