Viewpoint Based Adaptive Streaming For Panoramic Video

With the rapid development of virtual reality and the growing needs for immersive and highly interactive new multimedia,the demand for online real-time panoramic video is growing.However,the high bandwidth,low delay requirement of panoramic video brings great challenges to the development of its online service.On one hand,most of the current online video service is difficult to ensure both real-time viewing and the quality of service,since most of the videos are transmitted with traditional streaming technologies.On the other hand,the actual visual part of the user is only the current visual angle of a panoramic video surrounded by a panoramic view,while the information outside the viewing angle occupies a large amount of network bandwidth resources,causing great waste to the network resources.In this paper,a viewpoint-based adaptive transmission scheme for panoramic video is proposed to meet the users' demands for online services under current network conditions.The panoramic video is encoded into different quality versions,each divided into multiple tiles,and each tile is delivered to the client using parallel DASH streaming.With the combination of multi-tile stream and DASH technology,the whole service is dynamic and adaptive in two dimensions of time and space.This scheme dynamically adjusts the transmission priority and quality according to the current network conditions and viewpoint feedbacks.The visible part is given higher priority to ensure its quality and the bitrate in the invisible part is reduced,accordingly.The simulation results show that the adaptive transmission of panoramic video based on viewpoint and network environment can effectively alleviate the contradiction between the network demands and the current network conditions.The proposed scheme can save up to 60%of the bandwidth resources and provide users with better quality of service than the traditional scheme of transmitting the full panoramic video under the same network conditions.