Rule-based combination of video quality metrics

Lately, several algorithms have been proposed to automatically estimate the quality of video sequences, even some have been included in international standards. However, the majority only provide high performance under particular conditions and with certain types of degradations. Therefore, some proposals have been presented setting out the combination of various quality metrics to improve the performance and the range of application. In this paper, a rule-based combination of standardized metrics is presented, in contrast to most of these type of approaches based on combinational models. The proposed system consists of a first stage in which the type of degradation affecting the video quality is identified to be caused by coding impairments or transmission errors. Then, the most appropriate metric for that distortion is applied. Specifically, VQM and VQuad have been considered for coding and transmission distortions, respectively. The results show that the overall performance is better than using the quality metrics individually.