Research On Image Scrambling Algorithm Based On Segmentation

With the rapid development of Internet technologies, and the increasing popularization of the network, more and more information widely spread online, the network has gradually become the main tool of human information communication. The image information as the most important information of multimedia has been more and more popular on the internet, the problems of information security also occur at the same time. So when sending images online, the owner of the images must make reliable encryption for processing the images to prevent theft or tampering. As a result of that the image encryption and the hiding techniques have come into being, and have been widely applied in government, national-defense, banking, privacy and many other areas, and also play a more and more important role.Because the image has a large amount of data information, high redundancy, a strong correlation between the pixels, the traditional encryption algorithms such as DES, AES are no longer suitable for image encryption. The encryption algorithms must be considered to the characteristics of the image, the general algorithms scramble in the space domain, in frequency domain, or both. These algorithms have simple operation and no data expansion, reduce the correlation between pixels, and achieve image encryption. If one does not know the scrambling algorithms, it is very difficult for the eavesdropper restore the original image by the brute-force methods.At present, there are several typical image scrambling algorithms such as Arnold transform, Affine transform, Magic transform, Backer transform, Hilbert curve, Cray code, etc. But these scrambling algorithms have many defects such as inherent cyclical, small key space, no strict separation between the encryption algorithms and the key, encryption algorithms can not be released, all these defeats mentioned above can not meet the requirement of the modern cryptography.In order to overcome these problems, the works of this thesis are as follows:① The studies are focused on several typical scrambling algorithms, including the main idea of encryption and decryption, as well as the advantages and disadvantages.② An image scrambling algorithm based on the macro-segmentation is proposed, which has improved deficiencies for the several typical scrambling algorithms. A chaotic equation is introduced as key generation algorithm, furthermore, the encryption algorithms itself and the key can be separated, the details of the encryption algorithms can also be released. Following the Kerckhoffs’ Principle, and the requirements of the modern cryptography are satisfied.③ An image scrambling algorithm based on the micro-segmentation is proposed, which achieves synchronous scrambling in the space domain and frequency domain and speeds up scrambling. The values of the adjacent pixels are computed iteratively, such that the adjacent pixels have stronger dependence, moreover the proposed algorithm has also increased the sensitivity of the plain-text image, and effectively resisted the differential attack.④ To verify the scrambling effect and security performance, a series of theoretical analyses and experimental simulations are done through several typical algorithms and the proposed algorithms by combing with the evaluation criteria of image encryption algorithm. The results also illustrate that the proposed algorithm has better scrambling effect and security performance.