Tiling and Downsampling-based Immersive and Multiview Video Streaming Systems

Panorama photos/videos provide a brand new way to demonstrate 360 degree view of a scene in different platforms. While existing techniques can create and store the photos/videos as a whole frame, the rest part of this thesis presents a development of efficient tile-based coding system for storing and transmitting the panoramic photos/videos, thus preserving the bandwidth and improving the quality of the photos/videos. It also improves the user experience especially for mobile users. Furthermore, the latest coding standard high efficiency video coding standard (HEVC) and its extension is utilized to replace H.264/AVC coding structure in original tile-based system. With a model-based rate distortion optimization algorithm for choosing the quantization parameters across different tiles and layers, we further improve the rate-distortion performance for this system. In the second part of the thesis, we focus on improving the efficiency of interactive multiview video streaming system, where users can switch view points during the playback, and the system can generate virtual view. A novel downsampling-based interactive multiview video stream system is proposed to offer a new degree of freedom to the system design. With the bitrate adaptive downsampling ratio selection and joint depth-texture bit allocation, the rate-distortion performance of the system can be improved at low rate regimes.