A novel approach for characterizing multimedia 3D video streams by means of quasiperiodic processes

This work proposes a novel methodology to describe the low-frequency behaviour of compressed 3D video streams, i.e., their average fluctuations on longer timescales. This study is innovative for two reasons. First, it proves that the low-frequency behaviour of the video data belongs to the class of quasiperiodic processes. Second, it proposes an innovative approach to describe the long-term behaviour through a set of parameters directly derived from the quasiperiodic analysis. Reported results show that the proposed approach is effective in a wide variety of simulation scenarios. Furthermore, it can be easily generalized to other kinds of compressed two-dimensional (2D) streams, whatever be the adopted algorithm, the compression degree, video resolution and format. This opens new unexplored possibilities in the field of 3D video characterization, identification and classification.