Font Size: a A A

Detecting Of Video Steganography For Network Media Content Supervision

Posted on:2015-10-30Degree:MasterType:Thesis
Country:ChinaCandidate:G S ShiFull Text:PDF
GTID:2298330467972359Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
With the rapid development of digital multimedia technology and internet techniques, especiallythe widely used of video site, the information security has been more and more necessary. Themanagement of media contents in network needs to monitor or filter harmful information andprovide valuable management and analysis tools for government propaganda departments andnetwork regulatory department. Video information hiding has emerged to be an active andflourishing research topic in the field of information security recently. As the contrary, videosteganalysis plays an important role in information security. However, there are few research basedon the use of video steganalysis to manage network media content in recent years. In this thesis,video steganalysis was discussed. The main contributions can be summarized as follows:This thesis firstly studied the principles of video compression and proposed that video codectechnology is the base of video steganography and steganalysis techniques on compressed videos.Meanwhile, this thesis analysised the principles and characteristics of video steganographytechnology based on the study of video steganography and steganalysis techniques. Then, videosteganography and steganalysis techniques were classified and summarized.Our work targets werebased on the analysis.Secondly, this thesis makes improvements of the existing motion vectors steganalysisalgorithms for the low detection rate and high computational complexity of current steganalysisalgorithms by proposes the detection algorithm which use mutual information and combiningtemporal and spatial correlation to extract feature vectors. Mutual information and the mean,variance, skewness, kurtosis of the difference histogram of predicted frames ware used as thecharacteristics of inter-frames and the entropy of a binary data stream as the characteristic ofintra-frames. The support vector machine was used as classifier. Experimental results show that thealgorithm is easier to implement and better in detection performance than other methods.Thirdly, a blind detection method which is based on feature fusion aimed at the compressedvideo is proposed in this thesis. Firstly, steganalysis feature library which contains many validatedfeatures is been established, and there may be correlation and redundancy between multiple features,feature fusion and feature selection are used for each feature sub-libraries to reduce the number offeatures to improve the classification results. Video stream using steganography algorithm based onDCT coefficients and motion vectors (MV) are used in experiments. Experimental results show that the algorithm can effectively detect these video steganography techniques.Finally, this thesis describes network media content management system and its necessity. Theframework of network media content management based on the technology of video 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.
Keywords/Search Tags:video steganalysis, motion vectors, spatial correlation, temporal correlation, SVM, feature fusion
PDF Full Text Request
Related items