Privacy Protection Oriented Image Steganography

With the development of big data and cloud computing, demand of information security and data privacy in cloud environment is increasingly strong. In recent years, the leakage and abuse of privacy data make people worry about information security furtherly. Naturally, as one of the important research contents of Information Security, steganography techniques are receiving much attention today. Aiming at hiding existence of information within innocuous carriers, the goal of steganography is to modify the carrier in an imperceptible way, so that neither the embedding of a message nor the embedded message itself is revealed. For carring data, all kinds of multimedia can be used, such as text, voice, images and video. Among them, because of the wide application and large redundant space, image is the first choice. As early as half a year before the "911incident" occurred, the news is reported that digital images in some public web site are used to hide important information about the terrorist attacks by the terrorists. Last year, a former employee of CIA, Edward Snowden, exposed "Prism". Later, a series of photos uploaded by Snowden’s girlfriend caused wide public concern and discussion about image steganograpy. With the development of network and multimedia technology, more and more information is saved or transmitted in the form of images. Image steganography and steganalysis has become an important area of scientific research.In cloud environment, strong demand of information security and data privacy not only has driven the wide application of encryption algorithm and the development of ciphertext signal processing but also put forward new challenge to image steganography. Privacy protection oriented image steganography is not only a multidisciplinary cross research point concerned with Information Hiding, Cryptography, and Signal Processing and so on, but also is an emerging research hotspot which is driven by strong demand of privacy protection for now.On the topic of "Privacy Protection oriented Image Steganography", this dissertation has summarized the steganographic techniques research history and status quo and analyzed the application requirements for privacy protection and existing problems of published work. Our research is mainly focused on Reversible steganography research in Encrypted Images(RSEI) and Large capacity steganography in image spatial domain. The main work and innovation points are as follows. 1) Reversible Steganography research in Encrypted Images(RSEI)The demand for security and privacy of cloud data has promoted the rise of data encryption in clould computing and signal processing on ciphertext domain. That also puts forward new challenge to reversible steganography:Most of the reversible steganography methods in the plaintext domain do not work in the cipher text because of the changing of data features. Focusing on Reversible Steganography research in Encrypted Images(RSEI), the main research work and achievements are as follows.A. The innovation of most published RSEI methods appears in the part of reversible steganograpy rather than image encryption. Even a single image encryption method, simple stream cipher encryption using exclusive or operation, is adopted in the vast majority of existing RSEI algorithms. To solve this problem, Multi-granularity Encryption for Image(MGE) is proposed. Firstly, the cover image is divided into non-overlap blocks. Then, multi-granularity encryption is adopted by using random permutation:pixels permutation in each block and blocks permutation in the whole cover image. The experimental results show that, MGE is simple and has good encryption effect. More important, MGE can keep the local feature and histogram statistical properties unchanged.B. The secret data can only be extracted after image decryption in some published RSEI methods. If a receiver has the data-hiding key but no the decryption key, he cannot extract any information. In other words, data extraction is bound to threaten the privacy of image content. To overcome this problem, a separable reversible data hiding scheme in encrypted image based on multi-granularity encryption and residual histogram shifting is proposed. Compared with the existing methods, the scheme has the following advantages:1) simpler calculation and higher efficiency;2) larger embedding payload and better embedding quality;3) error-free recovery with increasing payload.C. Base on MGE, we propose another improved work:After a cover image being encrypted with a content-owner key, additional data can be embedded into the encrypted image based on block sorting and block histogram shifting with a data-hiding key. The experimental results demonstrate that the proposed algorithm has improved image quality while maintaining zero error rate. 2) Large capacity steganography in image spatial domainPrevious image steganography methods often pay more attantion to image quality rather than embedding capacity. Thus they can not meet the application requirements of prevacy protection under the background of big data. To solve this issue, focusing on large capacity steganography in image spatial domain, the main research work and achievements are as follows.D. A second order steganographic method based on modification direction exploiting(MDE) and adaptive reference matrix is proposed. The experimental results demonstrate the improvements with our method in terms of capacity, efficiency and detection rate compared to recent MDE-based methods.E. A data hiding algorithm based on pixel pair matching with large payload and acceptable visual effect is proposed. We use a group of functions to guide cover pixel pairs’ modification and imply secret digits. The experimental results demonstrate that the proposed algorithm has much larger embedding capacity than that of the published works, while keeping good invisibility in Human Visual System(HVS).