Design of video quality metrics with multi-way data analysis

Video quality metrics determine the visual quality of distorted video sequences by using prediction models based on objectively measurable features and are therefore an alternative to the time-consuming and costly subjective video quality assessment. In the conventional design approach to video quality metrics, however, the temporal nature of video is often considered only inadequately due to the use of temporal pooling in the prediction process. Moreover, this approach also often requires knowledge about the human visual system that is not readily or only partly available. In this thesis, I therefore propose a data driven design methodology using multi-way data analysis for the design of video quality metrics. This data driven design approach not only requires no detailed knowledge of the human visual system, but also allows for a proper consideration of the temporal nature of video by using a three-way prediction model, corresponding to the three-way structure of video. Using two simple example metrics, I demonstrate that this purely data driven approach not only outperforms video quality metrics in the state-of-the-art that are often highly optimised towards specific properties of the human visual system, but also that multi-way data analysis methods outperform the combination of two-way data analysis methods and temporal pooling.