MPEG-2 Prediction Residue Analysis

TL;DR

This paper develops a semi-analytic model to analyze the evolution of prediction residues in MPEG-2 double compressed videos, providing insights into the Variation of Prediction Footprint and its impact on forensic detection methods.

Contribution

It introduces a novel semi-analytic model based on synthetic signals and autoregressive models to explain prediction residue behavior in MPEG-2 double compression.

Findings

VPF behavior depends on scalar quantizer deadzone width

Model explains evolution of prediction residue in double compressed videos

Performance of forensic techniques varies with quantization parameters

Abstract

Based on the use of synthetic signals and autoregressive models to characterize temporal dependencies and inter prediction processes, a semi-analytic model is built to explain the evolution of the prediction residue in MPEG-2 double compressed video sequences. This characterization provides valuable insights on the behavior of the Variation of Prediction Footprint (VPF) exploited by different methods in the video forensics field, and also shows that the performance of the VPF-based techniques for GOP size estimation and double compression detection depend on the deadzone width of the scalar quantizers used for encoding each type of MacroBlock (MB).