First-order Statistical Characteristics To Maintain The Steganography Algorithm Research

Information hiding is an interlaced subject, which is built on the basis of mathematics, modern cryptography, information processing theory, and computer applied techniques. In 1990's, with the fast developing of network communication technology and multimedia technology, information hiding sprung up rapidly. Recently it has been a hot interest area in the research of information security and widely applied to such fields as computer communication, copyright protection of digital media, multimedia retrieval, and so on. Steganography and digital watermarking are two main sub-disciplines of information hiding.As the research of steganography became in-depth, the attack techniques against steganography called steganalysis came into being. The main purpose of steganalysis is to detect, extract and recover the hidden messages, which has challenged the security of steganography directly. So the capability of avoiding dectection has been a requirement to ensure the security of steganography system and the steganalysis-resisting algorithm has become the emphasis of steganography research.JPEG2000 is the new wavelet-based coding standard of still image made by Joint Picture Expert Group (JPEG) of the ISO. JPEG2000 has more outstanding performance compared with the widely used JPEG standard. Developing new standard oriented steganographic approaches is an inevitable trend of the research of steganograhy.This thesis is focused on first-order statistical properties preserved steganographic techniques, mainly including the following two facets:1. A spatial steganographic algorithm based on histogram modification is proposed. By analyzing the mathematical model of the process of random message sequence embedding, the statistical properties of the stego-image are derived, according to which the histogram of the cover image is modified to reach the goal of preserving the first-order statistical properties of the cover image after the embedding process. Experimertal results show that this new algorithm achieves good performance of preserving the first-order statistical properties and those steganalysis techniques against LSB-based algorithm cannot detect the existence of hidden messages.2. A wavelet-based steganographic approach is proposed. On the basis of analyzing the capacity-distortion model in the wavelet domain, a modified wavelet-based statistical properties preserved staganographic method is proposed. The application framework of this approach in JPEG2000 is also given. According to the analogical experimental results, this approach has good properties of avoiding detection.We also analyze some existing steganographic algorithms, steganalysis algorithms andthose corresponding modified steganographic approaches.Finally we summarize our research work, and discuss the further research topics and directions in the future.