A Stereo Video Quality Assessment Method for Compression Distortion

Based on feature analysis of video compression distortion, a new stereo video quality assessment (SVQA) method for compression distortion is proposed from temporal characteristics of video and binocular perception in this paper. First of all, significant pixels of monocular video are determined, and a new definition of region-of-interest (ROI) is given by using significant pixels and just noticeable distortion model. Secondly, taking ROIs and temporal perception characteristics of video and depth perception of stereo views into account, two quality metrics, left and right views´ quality metric and depth perception metric, are designed. Finally, the two quality metrics are pooled to obtain final SVQA score. Experimental results with tested in the NAMA3DS1 - CoSpaD1 stereo video database show that the proposed method is quite efficient for evaluating stereo video quality, its linear correlation coefficient (LCC) is above 0.92, and its evaluating results are more consistent with subjective perception.