Time Series Analysis of H.264 Video

Abstract

The aim of this paper is to investigate the statistical properties of H.264/AVC video sequences. The stationarity, dependence properties and self-similarity are investigated. Analysis of the H.264/AVC video sequences variable bit rate (VBR) traffic has shown that this data has strong self-similarity properties as well as noticeable long range dependence (LRD) and short range dependence (SRD). The stationarity of H.264 video sequences is investigated by applying the fractional differencing technique and implemented with Haar maximal overlap discrete wavelet transform. The results indicate that the H.264 video is 2nd order stationary.