Abstract
The proliferation of three-dimensional (3D) video technology increases the demand for multiview video (MVV) communication tremendously. Applications that involve MVV constitute the next step in 3D media technology, given that they offer a more realistic viewing experience. The distribution of MVV to users brings significant challenges due to the large volume of data involved and the inherent limitations imposed by the communication protocols. As the number of views increases, current systems will struggle to meet the demand of delivering the MVV at a consistent quality level. To this end, this thesis addresses efficient coding, adaptive streaming, and loss-resilient delivery techniques for MVV. The first contribution of this thesis addresses the problem of cost-efficient transmission of MVV with a provided per-view depth map. The primary goal is to facilitate the delivery of the maximum possible number of MVV streams over the Internet in order to ensure that the MVV reconstruction quality is maximised. Accordingly, a novel view scalable MVV coding approach is introduced, which includes a new view discarding and reconstruction algorithm. The results of extensive experiments demonstrate that the proposed MVV coding scheme has considerably improved rate-distortion (R-D) performance compared to the state-of-the-art standards. The second contribution of this thesis is the design of an adaptive MVV streaming technique that offers uninterrupted high-quality delivery to users. In order to achieve this, a number of novel mechanisms are introduced that can adapt the MVV content to collaborative peer-to-peer (P2P) and server-client multimedia dissemination networks. Experiment results show that the suggested adaptation technique yields a superior playback performance over a broad range of network conditions. The final contribution of this thesis is the design of an error-resilient scheme that addresses packet losses for MVV streaming. The aim is to make the MVV streaming more reliable against communication failures. Simulation results clearly show that the proposed approach outperforms reference solutions by a significant margin, not only objectively, but through subjective testing as well.