Research On No-Reference Video Quality Assessment Method Based On H.264

Recently, video sharing websites, IPTV, mobile TV, video surveillance and other video applications emerge in large numbers. As the most advanced video compression technology and its excellent network compatibility, H.264 is widely used in these areas. Because of image distortion introduced in video data collection, compression, transmission and recovery process, video service providers urgently need a real-time video quality assessment method to take measures timely for protecting the QOS(Quality of Service) when video quality becoming bad. Non-reference video quality assessment did not have the good assessment model at present. The existing assessment models often have many assumptions, limited application scenes, inconsistent with the subjective assessment and so on questions.In this paper, we research on the non-reference video quality assessment method for H.264 stream. First, we analyse the video quality injuries from reasons of instability network and video compression. Then we propose a network quality autoadapted H.264 video quality assessment model after analyzed of existing models. This assessment method monitors known network parameters which influence video quality and judge network conditions of H.264 stream at first. We believe that video compression is the main reason for video quality injuries when network quality is well and network transmission is the main reason when network quality is poor. Then, according to two kinds of damage video's characteristics, we used different quality assessment methods for the video of two kind of network quality.Experimental test results show that this model has the good uniformity to video quality's assessment and the subjective assessment while satisfies timely assessments video quality.