Font Size: a A A

Research On Delivery Of Panoramic Video Based On Viewpoint Prediction

Posted on:2020-02-05Degree:MasterType:Thesis
Country:ChinaCandidate:Z Q WangFull Text:PDF
GTID:2428330590971655Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of multimedia technology,virtual reality technology has attracted more and more attention.Panoramic video service as part of virtual reality technology has also ushered in an explosive growth.However,panoramic video has a large amount of source file data,which is a huge challenge to the transmission network.Therefore,how to design transmission mechanism scientifically and reasonably to coordinate the relationship between external environment and panoramic video quality on the basis of guaranteeing user experience has become an important research to break through the bottleneck of panoramic video development.With the rapid development of multimedia technology,virtual reality technology has attracted more and more attention.Panoramic video service as part of virtual reality technology has also ushered in an explosive growth.However,panoramic video has a large amount of source file data,which is a huge challenge to the transmission network.Therefore,how to design transmission mechanism scientifically and reasonably to coordinate the relationship between external environment and panoramic video quality on the basis of guaranteeing user experience has become an important research to break through the bottleneck of panoramic video development.This article carries out further research on the basis of the predecessors,and the main contents are as follows:High quality of video content and high viewport prediction accuracy were demanded for viewport based adaptive panoramic video transmission,however the existing viewport prediction method had poor prediction accuracy,especially for long-time viewport prediction.In this paper,a scheme of viewport prediction for virtual reality based on the similarity of user motion behavior was proposed.First,linear regression model was adopted for preliminary viewport prediction based on the historical trajectory of the user's head movement.Then,based on the similar characteristics of the user's motion behavior,similarity model was used to analyze and predict the user's viewport.Finally,the prediction results of two models were jointly considered for viewport prediction.Accordingly,the viewport probability distribution on each video tile was calculated.Moreover,according to the results of the viewport prediction,the QoE(Quality of Experience)based tile representation selection method for adaptive panoramic video transmission was designed.The experimental results showthat the proposed prediction method based on similarity of user motion has better performance in viewport prediction accuracy.At the same time,the proposed tile representation selection method could achieve better QoE,which also has high practical value.When multiple users subscribed to the same video content at the same time in the same area,multicast technology was a more appropriate transmission technology.However,most of the current multimedia multicast transmission systems focused on the allocation of resources in the transmission network system,while ignoring the viewpoint characteristics of panoramic video and the difference of video resolution between different regions,which seriously affected the application of panoramic video in practice.To solve this problem,a packet multicast panoramic video transmission mode was proposed.According to the user's viewpoint information,channel conditions and available resources in the transmission network,the appropriate bit rate module was selected for transmission to system users.In this model,firstly,according to the panoramic video viewpoint prediction mode based on user's similar mobile behavior,the range of the viewpoint area of a particular user and the corresponding probability values of each Tile were obtained.Then,according to the probability values of Tile,the transmission weight was determined.Finally,according to the user's viewpoint information,channel conditions and the available resources in the transmission network,the transmission weight was determined.In order to maximize the average transmission bits of panoramic video streams,users were divided into multicast subgroups and the transmission bits of Tiles in each subgroup were determined.The experimental results show that the proposed packet multicast panoramic video transmission mode effectively improves the video quality in the user's viewpoint area while taking into account user fairness.
Keywords/Search Tags:virtual reality, panoramic video, viewpoint prediction, multicast
PDF Full Text Request
Related items