Perceptual quality assessment of 3D videos with stereoscopic degradations

Abstract

In the last decade, several aspects of the 3D video technology have been improved, including the 3D content production, distribution, and display. Still, the level of acceptability and popularity of 3D video applications are strongly correlated to the user Quality of Experience (QoE). Since research in this area depends heavily on data acquired in psychophysical experiments, public databases with typical stereoscopic degradations are considered important tools to researchers. Although currently there are number of available public 3D video quality databases, most of them contain only compression and transmission degradations. In this work, our first goal was to build a database (UnB-3D) that contains stereoscopic distortions. We created a set of Computer Graphics Imaging (CGI) scenes and rendered it using different parameters, generating 3D videos containing stereoscopic degradations at different strengths. Our second goal is to understand how these stereoscopic degradations are perceived by viewers. So, we performed a psychophysical experiment to analyze the perceived quality and comfort of these videos. Finally we conducted the statistical analysis and model generation. Results shows that users that have little familiarity with 3D content have difficulties identifying stereoscopic distortions. Also, the source content has a great influence on the user’s comfort. Similarly, the 3D quality is affected by the spatial and temporal information of the content, specially when the disparity is high.