الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 128 |  |  |
| | | from | 128 | | |
Abstract
Scientific research currently places much focus on image encryption and steganography techniques, since they have proven their importance and effectiveness in the field of multimedia communication systems. Researchers have worked hard and have succeeded in reaching different algorithms to perform such tasks, however, there is a continuous need for more advanced algorithms with better security and more resistance to attacks. This thesis presented new image encryption and steganography techniques that are more efficient and highly secure. This research proposed an image encryption algorithm of high security and high efficiency. The properties of DNA were used as a guiding measure in the application of these new techniques, along with DNA’s complementary and dynamical range rules, as well as the complex and nonlinear Chen’s hyper-chaotic map. All these methods were used for the construction of the encrypted image. This newly developed encryption algorithm was compared to other more conventional ones and yielded better results. Algorithms used for comparison included those based on the properties of DNA, along with different chaotic maps, and double random phase encoding. Another inspiration, which was applied in this research, was the use of pseudorandom number sequences based on fuzzy integrals. These sequences are dependent on two-component keys. These two components are a pseudorandom secret sequence of limited length and a secret image. The latter represents the new concept presented in this thesis and can be a stamp, a signature, etc. Complexity and non-linearity are among the characteristics of fuzzy integrals, which makes for pseudorandom sequences of high security. This thesis considered the Choquet fuzzy integral, a special type of fuzzy integrals. Positive results have been obtained from testing the generated sequences, using the National Institute Standard Technology random test package. When generating these fuzzy pseudorandom sequences, researchers should be very mindful of the fact that the generation process is very sensitive to any changes in the pseudorandom or image key. An uncorrelated sequence that is totally different is the result of a single-bit change in the pseudorandom key. Thus, this thesis proposed a new more efficient technique to be used in image encryption and steganography. This technique employs, for the first time, methods that include hybrid DNA encoding and Choquet fuzzy integral sequences, to encrypt images. For an additional layer of security, this technique uses a new method of steganography, in which the image to be encrypted gets divided into four sub-images. A given key is used to choose different carrier images, in which each of these sub-images are hidden. The efficiency of the algorithm, suggested in this research, for image encryption was proven through simulation and a security analysis, which also confirmed that the steganography approach provided better security. Higher key sensitivity was also proven for the suggested algorithm, through the results of the tests performed in this research. Minuscule values for the pixel correlation coefficient were obtained. They were between 5.3220e-04 and 0.0011 horizontally, between 8.7670e-04 and 0.0022 vertically, and between 0.0002 and 0.0045 diagonally. Values between 7.9970 and 7.9979 (close to the ideal value of 8) were obtained for information entropy of the images being encrypted. The measured unified average changing intensity was in the range of 33.46‒33.39. As for the number of pixel change rate, its values fell within the range of 99.61‒99.64. Both ranges were close to the ideal values. The steganography test was used to check the visibility of the encrypted images, which were almost invisible at high values of signal to noise ratio. When subjected to different types of attacks, their normalized correlation coefficient values were good. Thus, this thesis was able to prove the superiority of the newly proposed algorithm over other older ones used for the same purpose. The final part of this thesis proposed a novel image encryption algorithm, which uses Choquet fuzzy integral and DNA techniques, to generate four S-boxes used in encryption. This is the first time these S-boxes, named DNAFZ S-boxes, were introduced. To generate these S-boxes and ensure they are highly secure and random, a secret image and an external key are used. Additionally, schemes used in image encryption are based on hyper-chaotic systems and their dynamic features. The suggested DNAFZ S-boxes had superior statistical properties, under majority logic criteria. These criteria include contrast, correlation, entropy, homogeneity, and energy. To test the ability of the encrypted images to withstand different types of attacks, numerical simulation was employed. The pixel correlation coefficient was between 7.8597e-04 and 0.00527 horizontally, between 8.7856e-04 and 0.00452 vertically, and between 0.00241 and 0.00021 diagonally. All values were quite small. Values of the information entropy for encrypted images were very close to 8, the ideal value, as they were within the range of 7.9965:7.9989. The unified average changing intensity and the number of pixel change rate had values in the ranges of 33.46‒ 33.32 and 99.58‒99.62, respectively. Again, both ranges were close to the optimum values. All previously stated results proved the superiority in performance of the newly proposed algorithm compared to more conventional methods, which include the cubic S-box, the LSS chaotic map, the Dynamic Henon map, Arnold transforms, and hybrid chaotic map optimized substitution. In conclusion, the newly proposed algorithm, after showing promising results proving its robustness, is suggested for use in digital image encryption in current and improved future multimedia communication systems.