| In recent years,with the rapid development of multimedia and computer network technologies,digital images,the most intuitive information carrier,have become an important way for people to exchange information.More and more digital images are distributed over open networks,including the Internet and wireless networks,greatly facilitating the accessing and sharing of information.Meanwhile,image information in many sensitive areas such as commerce,finance,medical treatment,scientific research,military affairs,and politics faces huge potential security risks when transmitting under open network environments.It often encounters various kinds of potential attacks,including illegal copy,falsification,and distribution,which have caused huge losses to data owners.Therefore,the study of image encryption technology is of great significance.In recent years,the development of chaos theory has provided a new direction for image encryption technology.Since the mid-1990s,many scholars have noticed that there is a natural connection between chaos and cryptography.The intrinsic properties of chaos,including extreme sensitivity to initial conditions and system parameters,ergodicity,long-term unpredictability,and pseudo-randomness,makes it well satisfy the requirements of constructing a cryptographic system with good security performance.The encryption system constructed based on chaotic dynamics provides a good balance between security and encryption efficiency.It not only provides a high level of security,but also has the advantages of simple hardware and software implementation and high execution speed.Therefore,the chaos-based image encryption technology is especially suitable for the application of real-time and secure transmission of digital images,which are characterized by bulk data capacity.At present,the chaotic image encryption algorithm has become the mainstream technology and research focus of image information security,and has great potential for application.Based on the thorough analysis of the drawbacks encountered by the existing chaotic image encryption algorithms,this thesis proposes four novel chaotic image encryption algorithms using the multi keystream elements quantification method,the plaintext-dependent keystream generation method,the parallel image encryption method,and the one-time-pad encryption method,respectively.The innovation of this thesis can be summarized as the following four aspects:(1)Aiming at the problem of high computational cost for generating keystreams by iterating a chaotic system,a multi keystream elements quantization strategy is proposed,and the feasibility of the strategy is demonstrated by SP800-22 experiments.Based on this strategy,a fast color image encryption scheme is construcetd.In this scheme,the"scrambling" and "substitution" keystreams are generated from the hyper-chaotic Lu system and logistic map,respectively.By introducing a 24-bit quantization mechanism for substitution keystream generation,three keystream elements can be generated from one iteration.Compared with the traditional scheme,the number of iterations is reduced by 2/3,and the keystream generation efficiency is improved.The scrambling operation is implemented based on a sub-pixel swapping mechanism,which solves the drawbacks of the conventional area-preserving chaotic maps,such as the periodicity,fixed point,and image size limitation.(2)For plaintext-dependent keystream generation algorithms using an orbital perturbation mechanism,the keystrewam can not be reused among different substitution rounds.In order to solve this problem,a new algorithm for generating plaintext-dependent keystream is proposed.The algorithm dynamically associates the keystream elements with the plaintext during the pixel value mixing stage.As the plaintext associating operation has no relation with the chaotic system iteration process,the keystream can be shared among different encryption rounds,thereby improving the encryption efficiency.In the permutation stage,the pixle positions are scrambled using a Henon map-based pixel swapping mechanism.In the diffusion stage,the Lu system is employed to generate the keystream required for the substitution operation.In addition,the algorithm can effectively increase the diffusion intensity,thereby reducing the number of encryption rounds required to achieve the desired diffusion effect.(3)Based on the analysis of the mechanisms of the existing parallel chaotic image encryption algorithms,a new parallel image encryption framework based on bit-plane partition is proposed.The existing parallel encryption frameworks are based on the 2D space partitioning.Each thread processes a sub-image in the diffusion stage,and at the same time,it cooperates with inter-block scrambling or global scrambling to achieve a desired diffusion effect.They have no special requirements on scrambling algorithms and diffusion algorithms and have strong algorithm compatibility.The proposed bit-plane based parallel task decomposition strategy,in addition to being compatible with pixel-based diffusion algorithms,is also compatible with bit-plane-based diffusion algorithms.Experimental results show that the algorithm based on this strategy is approximately 5 times faster than the corresponding serial algorithm.(4)Based on the analysis of the mechanism and implementation efficiency of a typical diffusion algorithm,a "one-time-pad" image encryption scheme is constructed.The scheme consists of only a single diffusion part that uses a 4D Li hyperchaotic system to generate the keystream.The initial conditions of the chaotic system are determined by the SHA-224 Hash value of plain-image as well as the secret key.Due to the avalanche effect of the hash function,completely different keystreams will be generated for two slightly different images.As a result,desired diffusion effect can be achieved after only a single round of encryption,whereas at least two encryption rounds are required when employing the "permutation-diffusion" structure.In addition,the existing one-time-pad scheme does not consider the issue of key distribution and therefore has serious practical limitation.However,in the proposed algorithm,the hash value is not used as the secret key,and thereby can be transmitted together with the cipher-image in the form of plaintext.Therefore,the proposed algorithm has the advantage of low key distribution cost.Theoretical analysis and experimental results show that the above four encryption algorithms proposed in this thesis have high level security,and can be applied to the protection of confidential image data in military and commercial fields as well as personal private image data,thereby having good practical significance. |
PDF Full Text Request