Image Scrambling Algorithms Based On Knight-tour Transform And Evaluation Methods

The technology of digital image scrambling is an efficient supplement of image information hiding and it comes from the technology of encrypt. The scrambled image can't be recognized then the information is protected. The algorithm influences directly the security of digital image and other processing results. The scrambling algorithm based on knight-tour transform is a hot topic because the number of its secret key is large and its security is high. But it can only scramble the details of digital image. This thesis discusses the improved scrambling algorithms based on knight-tour transform to overcome the shortage. In addition, by analyzing and comparing many different methods that are used to evaluate the image scrambling result, two new mathematical formulas are proposed.The major work involved is as follows:1. An algorithm of knight-tour producing matrix is proposed. The image which is scrambled by the producing matrix can not only hide the large contour and edge but also have a whole unclear effect. The results of experiment indicate that it is feasible.2. A block multilayer scrambling algorithm is given. The size of block is decided by the size of image and there are three layers. The algorithm is not only easy to realize and has satisfied scrambling effect but also robust enough to some common image processing such as cropping, noise and JPEG compression etc.3. A composite scrambling algorithm is proposed which is made of knight-tour transform and Arnold transform. The image using the algorithm can be scrambled enough and its pixels distributing is uniform. The scrambled image and random noise image is indistinguishable. Encrypting and decrypting is fast.4. An improved formula based on Canny operator is proposed for accessing the image scrambling degree by analyzing the existing assessments. The method, known as scrambling degree, has little calculating amount, can reflect the scrambling result better from the image texture and its result is not affected by the original image. 5. A novel method is proposed for accessing the image scrambling degree, which utilities the favorable performance of DCT coefficients and the feature of masking of human vision system. The method correlates with subjective assessment and its results can reflect scrambling effect better of the same image and different images.