No-Reference Objective Quality Assessment for Video Communication Services Based on Feature Extraction

This paper proposes a no-reference video quality assessment method for multimedia communications and video services. Based on the analysis on the three types of transmission impairments, we define the feature parameters calculated at pixel domain. Then, the relationship between these parameters and the MSE calculated from some training videos is model. In practice of video quality assessment, the MSE indicating quality degradation degree is estimated from the receiver-side features parameters through the trained relationship model. Experimental results show that the proposed model provides satisfactory effectiveness on characterizing packet-loss impairments.