Universal Audio Steganalysis Based on Calibration and Reversed Frequency Resolution of Human Auditory System

摘要

Calibration and higher order statistics (HOS) are standard components of manyimage steganalysis systems. These techniques have not yet found adequateattention in audio steganalysis context. Specifically, most of current worksare either non-calibrated or only based on noise removal approach. This paperaims to fill these gaps by proposing a new set of calibrated features based onre-embedding technique. Additionally, we show that least significant bit (LSB)is the most sensitive bit-plane to data hiding algorithms and therefore it canbe employed as a universal embedding method. Furthermore, the proposed featuresare based on a model that has the maximum deviation from human auditory system(HAS), and therefore are more suitable for the purpose of steganalysis.Performance of the proposed method is evaluated on a wide range of data hidingalgorithms in both targeted and universal paradigms. Simulation results showthat the proposed method can detect the finest traces of data hiding algorithmsand in very low embedding rates. The system detects steghide at capacity of0.06 bit per symbol (BPS) with sensitivity of 98.6% (music) and 78.5% (speech).These figures are respectively 7.1% and 27.5% higher than state-of-the-artresults based on RMFCC.