Study On Data Hiding And Image Authentication Scheme In Encrypted Domain Based On Compressive Sensing

With the increasing of the network openness and the computing power, more and more digital images are transmitted and processed over the network, and are integrated into all aspects of daily life, work and study. Because digital image is easy to be modified and copied, when it is transmitted over an unsafe channel, there will be more and more security issues, such as malicious theft and tampering. The security of digital image gets more attention.The existing image encryption, information hiding and image authentication technology can effectively solve some information security problems. The data owner can encrypt data to hide information content. The hider can hide secret information into carrier information to achieve information protection and copyright protection. The receiver can ensure the integrity and authenticity of information. For some special fields, such as the military intelligence and medical, the recovery of the original image content is also necessary, but the existing research is ineffective. As the security and reconstruction features of compressive sensing(CS) become mature, more and more researchers begin to study information hiding and image authentication based on compressive sensing. In this thesis, through the systematical study on image encryption, information hiding and image authentication technology, the information hiding scheme and the image authentication scheme in cipher image based on compressive sensing are proposed.(1) Based on the existing information hiding algorithm and compressive sensing theory, a high quality image reconstruction and watermarking scheme in cipher domain is proposed. Through the analysis of image energy distribution characteristics, the original image is divided into important part and unimportant part. Then compressive sensing is used to process the unimportant part, and the watermark is embedded into the measurements. This will not only improve the embedding capacity, but also increase the security of the secret information. There the watermark location key will be generated according to the energy distribution of important part, which will not only be convenient for the positioning of watermark during embedding and extracting, but also improve the accuracy of extracted watermark. Through compressive sensing technology, we realize the encryption, compression and reconstruction of image. Experimental results show that, compared to the general information hiding algorithms, our algorithm has a strong watermark security and high quality image restoration, and can resist cropping attack, reduce the packet loss rate, so it can be used effectively in unsecure channel to transmit information(2) Through the study of existing image authentication algorithm and the reconstruction feature of compressive sensing, a encrypted image authentication and recovery algorithm based on compressive sensing is proposed. In the scheme, the original image is first divided into important part and unimportant part, and different encryption algorithms are processed for different parts. But as we want to achieve tamper verification and tamper recovery, so low-frequency part is processeed with compressive sensing to generate characteristic watermark which will be embedded in the unimportant part measurements. There the reconstruction characteristics of compressive sensing is utilized to implement tamper verification, tamper recovery. At the same time, the digest value is extracted from watermark as the tamper localization criteria. Experimental results show that the algorithm has better robustness and security. It not only ensures the full effect of encryption, but also can distinguish the type of tamper, localize tamper and restore important information of the image, which makes up the shortcoming of the existing studies.