Video Steganalysis Algorithm Based On A Spatial Temporal Detector

With the popularization of information technology and internet technology,the wide spread of multimedia resources have provided favorable condition for the development of information hiding technology.Information hiding has both advantages and disadvantages,one of its branch: digital watermarking can protect the copyright of the digital products,while another branch: information steganography provide convenience for illegal persons to transmit illegal information.So as its against technology,information steganalysis has emerged and become the research focus of information security.In recent years,because of the advantage of large embedded capacity,video has gradually become a favorable carrier for steganography,making the research of digital video steganalysis become more and more urgent.Spatial domain video steganography is the most widely used steganography method.However,the current spatial domain video steganalysis methods do not make full use of temporal and spatial correlation of video content and usually not suitable for actual steganalytic systems.This paper presents a video steganalysis scheme against the spatial domain video steganography based on a spatial temporal detector(ST_D)that considers spatial and temporal correlation of the video simultaneously.A 3-D orthogonal plane is established and three descriptors with noise-sensitive property are defined for XY,XT and YT planes respectively.Considering the influence of motion intensity and texture complexity of video sequences on these three descriptors,motion and texture analysis are proposed to divide each frame of the videos into different kinds of block.Then motion and texture factors are introduced to provide reasonable weights for histograms of the three descriptors of each block.This paper selects 14 standard CIF video sequences and carries out comparative experiments with three typical steganalysis model: SPAM,Rich Model and SPEAM for both uncompressed and compressed videos.The experimental results demonstrate the great advantages of ST_D,especially for compressed videos,the advantage is more obvious.