A Secure Steganography Preserving High Order Statistics

With the development of steganalysis, the security of steganography is faced with tremendous challenge. High order statistical analysis is an effective and common detecting method that can defeat many hiding scheme. This paper proposes a secure steganography that can effectively resist image steganalysis based on high order statistical analysis. We have analyzed Haar wavelet decomposition characteristics and obtained a theorem about keeping high order statistics invariability. Based on the theorem, the hiding method that can preserve high order statistics of H, V, and D sub-bands of images decomposed by Haar wavelet is devised. In order to strengthen the security of our steganographic scheme, the approach to generate the set of random embedding location is presented. Finally, the capability of proposed algorithm is discussed in theory and proved by experiments. The results show our method possesses perfect imperceptibility and can effectively resist high order statistics analysis and RS analysis.