Quality Assessment for Omnidirectional Video: A Spatio-Temporal Distortion Modeling Approach

Omnidirectional video, also known as 360-degree has become increasingly popular nowadays due to its ability provide immersive and interactive visual experiences. However, the ultra high resolution spherical observation space brought by large viewing range make omnidirectional video distinctly different from traditional 2D video. To date, quality assessment (VQA) for is still an open issue. The existing VQA metrics only consider spatial characteristics of distortions, but temporal change distortions can considerably influence human perception. In this paper, we propose a spatiotemporal modeling approach evaluate Firstly, construct spatioral unit average distortion in dimension at eye fixation level, based upon which smoothed value recursively calculated consolidated variations. Then, give detailed solution how integrate three into our approach. Besides, cross-format measurement investigated. Finally, whole sequence obtained pooling. Based on approach, full reference objective metric derived, namely OV-PSNR. experimental results show that proposed OV-PSNR greatly improves prediction performance