Research On Viewpoint Based Adaptive Streaming Technology For Panoramic Video

In recent years,virtual reality(VR)technology has developed rapidly,and panoramic video is an important business form in the field of VR.How to transmit high-bitrate panoramic videos and improve the quality of experience(QoE)under the network condition of limited and fluctuating resources is the current challenge.For a 360-degree panoramic video,the field of view(FOV)accounts for only a portion of the panoramic video,and the transmission of the non-FOV content wastes a lot of bandwidth resources.In order to solve the above problems,combined with Dynamic Adaptive Streaming over HTTP(DASH)technology,this thesis proposes a viewpoint-based adaptive panoramic video transmission scheme.This solution divides the panoramic video into different video tiles,adaptively transmits the tiles according to the user's viewpoint and network conditions.Among them,the adaptiveness includes two aspects.On the one hand,the viewpoint-based tile adaptive algorithm predicts future tiles in the FOV according to the changes of user's viewpoint.In this thesis,the linear regression model is used for viewpoint prediction,and the FOV-Tile ratio is adjusted according to the prediction error to improve the accuracy of FOV prediction.By reducing the amount of non-FOV-Tile data,more bandwidth resources are allocated to FOV-Tile,the problem of network capacity limitation in panoramic video transmission is solved.On the other hand,QoE-based bitrate adaptive algorithm dynamically selects bitrates for different tiles according to the changes of network.In this thesis,the QoE model of panoramic video,bandwidth prediction and real-time bitrate prediction are designed.The bitrate selection problem is converted into a QoE optimization problem,which solves the problem of network capacity limitation in panoramic video transmission.Finally,this thesis builds a panoramic video transmission simulation platform,and uses real users' head motion data and real network bandwidth data to emulate the user's viewpoint changes and the network changes.The experimental results show that,in the case of limited bandwidth,the proposed scheme can obtain higher QoE than the full transmission scheme.