A fusion approach to video quality assessment based on temporal decomposition

In this work, we decompose an input video clip into multiple smaller intervals, measure the quality of each interval separately, and apply a fusion approach to integrating these scores into a final one. To give more details, an input video clip is first decomposed into smaller units along the temporal domain, called the temporal decomposition units (TDUs). Next, for each TDU that consists of a small number of frames, we adopt a proper video quality metric (specifically, the MOVIE index in this work) to compute the quality scores of all frames and, based on the sociological findings, choose the worst scores of TDUs for data fusion. Finally, a regression approach is used to fuse selected worst scores from all TDUs to get the ultimate quality score of the input video as a whole. We conduct extensive experiments on the LIVE video database, and show that the proposed approach indeed improves MOVIE and is also competitive with other state-of-the-art video quality metrics.