Fractal and Multifractal Characterization of 3D Video Signals

Abstract

Video signals are responsible for the largest amount of information in data storage and data transmission systems. Three-dimensional (3D) video formats are the largest in terms of the amount of data, and thus the required bit rate. For the efficient transmission of 3D video in communication systems, a detailed knowledge of the traffic characteristics of the format is necessary. In this research, characterization of 3D video signals is performed by using fractal and multifractal analyses. Codes for analyses are written in MATLAB and Python. Communication network traffic shows self-similar behavior with long-range dependence (LRD). Using visual methods and a rigorous statistical methods, it was shown that video sequences of 3D video formats have fractal self-similar properties with LRD and high Hurst parameter values. It is shown that investigated 3D video formats have multifractal structure by using the histogram method. The research included different 3D video formats, specifically with one video sequence (frame compatible, FC and frame sequential, FS) and with two or more video sequences (multiview, MV). Cases with different signal qualities defined by the quantization parameter, different types of frames, different groups of picture (GoP), and different broadcasting methods were taken into account and separately analyzed.