Real-time Optimization of Free Viewpoint Television (FTV) System

OData support
Dr. Huszák Árpád
Department of Networked Systems and Services

In the last few years multimedia techniques development was accelerated. These days everyone can obtains cheap 3D or smart television or these combinations. However, the evolution does not stop. We can hear a lot about upgraded and new devices and services even now, such as the Free Viewpoint Video/Television. This technique’s essence is that the user can change his/her viewpoint freely in real time while sitting front of the television. We have to record the items or events with a lot of cameras at the same time, which viewing directions are different. Due to the large number of camera images the network traffic is increased significantly.

It makes difference, where the original images are converted to a virtual video for the client. If we do it in the lower layers of the network, the original (big size) images generate a huge traffic in the network, but if we convert it in higher levels, redundancies generate unnecessary traffic.

In my thesis I developed a software, which helps to optimize a Free Viewpoint Television/Video distributed network, so that the number of served clients are maximal and the whole traffic in the network is minimal. In this test environment we can set the network’s devices with its properties and the software examines the network with the selected method and gives the optional or nearly optional arrangement if it is exist or inform us about the errors if a solution is not acceptable. The optimmization process can run in real-time, so the network can adapt itself to continuously changes user needs.

I made several measurements with the developed simulation tool, where the network properties and the topologies were different. Compared the obtained results, I drew conclusion and proposed a solution to find the appropriate network layout.

A good layout can greatly reduce the network’s traffic so we can save up money and serve more users thus we can increase our income.


Please sign in to download the files of this thesis.