3D multimedia have been going through a constant and rapid development these years, resulting new media contents such as FVV (Free Viewpoint Video). FVV allows users to get access to an arbitrary viewpoint of the video content in real time. The scenes are usually recorded with several cameras so that later a viewpoint can be synthetized for any virtual location, where no real camera is deployed.. It’s clear, that transmitting more camera stream simultaneously could mean transmitting huge data.
In my thesis I am going to show algorithmic solutions on how to choose camera-images to be transmitted in order to reduce the number of camera handovers and avoid starvation of viewpoint synthesizer algorithm, while keeping the quality parameters on a certain level. To do this, I am also using results of viewpoint prediction methods. According to my simulation results, these algorithms reduce the forementioned values as expected. In cases of starvation 46% reduction was achieved, while using proper viewpoint prediction, the number of camera handovers were reduced by 26%. These results can be used for further analysis.