Detecting Of Audio Steganography For Network Media Content Supervision

With the development of computer networks and multimedia technology, and the rapid changesof international situation, information security technology has been growing attention. Informationhiding is one of the the technique in information security technology. At the same time, It is alsoused in the field of commercial secret communications, and also be used by criminals as a criminaltool. Steganography detection is the steganography against technology, can effectively monitor theillegal abuse of steganography, on the other hand, steganography detection technology also continueto promote the development of steganography. Steganography detection technology designed todetect whether the network media to hide the existence of bad information. Audio is one of severalmedia network carrier, so the study of audio steganography detection technology is of greatsignificance.Firstly, we studied one kind of method which can detect audio embedding secret information byusing Distortion compensated quantization index embedding. Distortion compensation quantizationsteganographic embedding method has better invisibility and robustness, this thesis draw ontraditional Quantization Index Modulation steganography detection algorithm to analyze the audiocharacteristics before and after embedding secret information changes, and draw its frequencydomain Histogram coefficient depending on the audio segment through discrete wavelet transformcoefficient difference before and after the change of characteristics is given a special steganographydetection method based on the difference between the amount of secret carriers coefficient variation.The experimental results and simulation results show that this detection algorithm is effective forboth audio distortion-compensated quantization index data hiding and audio simple QIM datahiding.Secondly, this article through the perspective of universal steganalysis, then, we have to throughseveral feature fusion methods to achieve audio detecting. However, this feature is usually fused tothe hidden information detection method discuss the problem as a deterministic study, ignoring theeffects of uncertainty in the study, which would lead to performance degradation steganographydetection. Based on this, the fourth chapter analysis of the common steganography detectionthrough uncertainties, then, a steganalysis model considering the uncertain factors from featuresextraction, classifier training and decision is put forward. Moreover,an universal steganalysisalgorithm is proposed based on Dempster-Shafer(D-S) evidence theory.The experimental resultsshow that the reliability and scalability ofthe algorithm is higher than previous algorithms anduncertainty theory is validity to resolve the steganalysis problems. Finally, this thesis describes network media content management system and its necessity. Theframework of network media content management based on the technology of audio steganalysis isproposed in this thesis. Using video steganalysis to build the regulatory mechanisms of networkmedia content management system, construct a green network environment through the use ofcontent filtering, blocking isolation and some other methods. By the analysis of the blocked mediacontent，this system can ensure network security and reliable communications.