Technical Research And System Implementation Of Adaptive Streaming Transmission For Panoramic Video

With the rapid development of virtual reality technology,the demand for immersive media such as panoramic video is increasing.However,high-resolution panoramic video has a high video bit rate,which brings great challenges to network transmission.The traditional video transmission scheme transmits the video completely on the network,and the user visible area in the panoramic video is only within the viewing angle range,and the video content in the background area outside the viewing angle wastes the network bandwidth resource.Existing research results currently adopt an adaptive transmission scheme based on the perspective,and classifies each region in the panoramic video content space according to the position of the viewing angle,and adopts a corresponding bit rate to transmit each video region according to the quality level.In this paper,the tile-based panoramic video transmission is studied and a viewport-based rate adaptive algorithm is proposed.The DASH(Dynamic Adaptive Streaming over HTTP)streaming technology is used to dynamically transmit video content.The architecture design and development implementation of the transmission system is also proposed.Viewport-adaptive streaming of panoramic video usually uses tile to divide the complete picture into multiple sub-areas.In order to measure the transmission quality of panoramic video,this paper constructs the QoE(Quality of Experience)model of view-adaptive streaming.The QoE model includes an objective function for evaluating the quality of the viewport picture and a constraint for rate control.The quality evaluation method measures the picture quality and stability of each tile video based on the designed viewport distortion and smoothness.The rate control method dynamically adjusts the upper limit rate of tile video set according to the bandwidth estimation and buffer level during transmission.Aiming at this QoE model,this paper further proposes a rate adaptive algorithm based on marginal utility optimization.According to the QoE marginal utility sorting sequence,the tile video is upgraded one by one,and the optimized bit rate selection is determined for the tile video set.The algorithm uses limited network bandwidth to enhance view QoE and improves user experience.Through comparison experiments,it is verified that the performance of the proposed algorithm is greatly improved compared with the traditional method,and it is also significantly improved compared with the existing adaptive method.At present,most of the research on adaptive video streaming of panoramic video is based on system simulation,which can not reflect the effect in the real environment.According to this situation,this paper designs and develops the panoramic video transmission system.The master-slave structure is adopted,and the client calculates the QoE marginal utility based on the viewport for each tile video set according to the proposed QoE model,executes the proposed rate adaptive algorithm,and adaptively acquires the multiple tile video stream according to the rate decision result.During the transmission process,the client downloads the video segments in parallel,decodes the data and renders them into a complete panoramic video image,which is presented in the user's head-mounted display to provide an immersive video experience.Server management maintains multi-rate version of tile video clips and video index files,and provides DASH streaming services.The server side adopts a high-performance asynchronous server model design,and uses the thread pool to process process events in the video stream transmission process,and has excellent concurrency capability.The system performance test results show that the implemented panoramic video transmission system can process multiple tile video streams in parallel with high efficiency,and realize rate adaptive according to the viewport.