Packet layer model of HEVC wireless video quality assessment

In order to realize the real-time monitoring video quality and improving the user experience. Paper proposes a packet layer video quality assessment model with the characteristics of the newest High Efficiency Video Coding (HEVC) standard. Firstly, packet loss rate(PLR) and encoding bits are obtained from packet-header's information and then using encoding bits to predict frame type. Secondly, analyzing the coding structure of the random access pattern of HEVC and getting different frame dependencies in pictures. Thirdly, we get characteristic parameters: PLR, percentage of damaged frames and percentage of different temporal classification frame which loses the packet. Finally, subjective scores and the parameters are trained by a feed-forward Back-Propagation neural network to establish an objective assessment models. Experimental results showed that model has a high correlation between subjective score and objective scores, the Pearson and Spearman coefficients were 0.95024 and 0.94749.