Audio Steganography Research And Analysis Techniques

Over the past ten years, among various information security techniques, a new sub-discipline called information hiding has drawn tremendous attention of academe and correlative area. Steganography and steganalysis the younger branch of information hiding are the important parts of information warfare in the era of the internet. Digital steganography is used to exchange secret messages by military units or intelligence agencies, or protect private or sensitive data by companies or individuals, or even hide their communications from law enforcement by the terrorist organizations. Steganalysis is the critical technique to prevent steganography from being abused, and is significant to national security and social stabilization.This thesis studies audio steganography and audio steganalysis which use audio as carriers. Two audio steganographic algorithms were designed in this thesis for high steganographic capacity and security goals after surveying the appropriate background literature. Two audio steganalytic methods were presented too of which one was to detect the hiding data embedded by currently known steganographic tools and the other was a universal audio steganalytic method. The main contributions of the thesis are as follows.1) Based on psychoacoustic model and perfect reconstruction ability of the wavelet packet transform, a self-adaptive, high capacity audio steganographic algorithm was proposed, and it can resisted the visual attack on audio spectrogram.2) After studying on the RS and Chi-square steganalysis, the conventional LSB steganography algorithm was improved upon with a novel LSB steganography algorithm that could effectively resist steganalysis based on above statistical analysis.3) With sensitive statistics features, a novel steganalysis scheme which correlated with steganographic software tools was designed. As shown by the experimental results, the proposed method could detect the existence of hidden messages, and also identify the tools used to hide them with high accuracy.4) Based on the strong ability of wavelet Multi-Resolution Analysis and the one-class SVM, with sensitive statistics features, a universal audio steganalysis scheme was presented. The experimental results showed this scheme was acceptable.