Research On Qoe Model In Video Services

Based on the development of video technology and network communication technology,video service industry has been developing rapidly.The requirement of users' quality of experience is getting higher and higher.High-definition,multi-screen,social and interactive capabilities have become users' preferred choice in video services.At the same time,the technology of video codec,terminal playback devices,and video service types are constantly developing and innovating.It is a crucial issue to improve users' quality of experience in various video services.It is of great research significance and practical significance to establish an effective quality of experience evaluation system.International Telecommunication Union has determined the concept of Quality of Experience(QoE)based on the Quality of Service(QoS).Some researchers and organizations have proposed a series of bitstream based methods to evaluate video quality.However,most of these methods are proposed or validated for only one specific codec that is not advanced than H.264/AVC.And the validations of these model are usually based on the experimental environment and data,which is not convinced in the real world application of video services.Besides,the evaluation of single quality indicator cannot reflect the overall quality of the experience from users' perspective.In practical video services,the tendency of multi-terminal,multi-scenario and multi-type brings more challenges to QoE evaluation.How to establish a more comprehensive QoE evaluation method and improve the generalization capability of the model has become an urgent problem to be solved.In this work,we firstly research the traditional non-adaptive video services.By studying existing models and the similarities between H.264/AVC and H.265/HEVC encoded video streams,a compatible bitstream based video quality assessment model for both codecs is proposed with terminal playback and video compression indicators taken into evaluation.The proposed model is validated to have good performance for both codecs.The model parameter extraction tool and algorithm tool are developed and published.Then,different video service type is studied in this work.For HTTP adaptive streaming(HAS),this paper present a QoE evaluation model based on ITU-T P.1203 model.Both traditional algorithms and machine learning method are used in this model to evaluate comprehensive quality degradation factors,including compression degradation,the influence of different quality level,adaptive quality changing degradation,stalling impairment and time related effect.Through validations,the feasibility and validity of the use of both parameters and machine learning method are proved.It is also proved that the proposed model has a good overall performance.