Study On Reversible Steganography Based On 4 Coefficient Histogram Shifting And Steganalysis Based On Inter-view Correlation

The information security problem of the information age has been paid more and more attention, steganography and steganalysis technology as an important branch of information security field, is one of the hotspots. Digital video has become a natural good carrier of steganography because of its widespread exists as the mass media, and the video based on H.264 compression standards has become the main way of video transmission for its excellent performance. At the same time, 3D video has been attached more and more importance because of its good visual experience, and it will be the focus of future development.The key idea of histogram shifting is to build a steep histogram, In response to the characteristics, a kind of reversible video steganography algorithm based on 4 coefficient histogram shifting is proposed. The basic idea is that select 4 quantized discrete cosine transform(QDCT) luminance coefficients to form a 4 dimensional vector,and according to the size of vector, 3 or 2 or 1 bits of information are embedded by modify one coefficient with ±1. The experimental results show that compared with some other algorithm, better capacity, video quality, embedding efficiency, bit rate increase, efficiency and safety can be got by using this algorithm.In response to the problem of the movement will have a great disturbance on steganalysis algorithm based on time correlation, a kind of 3D H.264 video steganalysis algorithm based on inter-view correlation is proposed. The basic idea is that the overlap region of 3D video's left and right view has a strong correlation, and the correlation will not change with video type, and the existing 3D H.264 video steganography algorithm embedded information mainly in the 4x4 transform block of right view. Compute the distributes difference of all 4x4 transform block in right view and the corresponding pixel in left view, use it to represent the inter-view correlation and detection. The experimental results show that, good detection effect can be got by using this algorithm.