Scrambling And Its Evaluation Method Of Study

With the development of the multimedia technology and the popularization of the internet, images become one of the most important ways to transmit information because of the visibility and abundance of information. However, some images are related to national security, business secrets, and personal privacy. Once they were stolen in the transmission, it will lead to serious consequences. So it is a challenge task that how to send these images to the receiver intact and safely. Image scrambling technology which can effect an image show of a random distribution of the pixels or similar to the characteristics of noise, can encrypt images perfectly, can also be used for digital watermarking, image hiding, image pre-processing sub-keeping and after the treatment process, and also an important way to solve the problem.This dissertation takes new technologies and methods of image scrambling and performance evaluation as its researched objects, and the main contributions are as follows.(1) An image scrambling method based on bit-exclusive-or is suggested. The method does a bit-exclusive-or operation on each pixel with its prior one first. Then the bits of each pixel value is exchanged across, thereby the effect of image scrambling is improved again. Experimental results show that compared with some existing algorithms such as the Arnold method, our method not only provides with a scrambled image whose histogram obviously changed, but also the scrambled image looks more like Gaussian white noise, which is promising when being applied into the image encryption transmission or the preprocessing of secret information in digital watermarking.(2) Another image scrambling method based on bit-exclusive-or operation is proposed. This method does a bit-exclusive-or processing on each pixel with its prior one first. Then the bits of each pixel value is exchanged across, thereby the effect of image scrambling is improved again. Experimental results show that compared with some existing algorithms such as the Arnold method, Hilbert method and chaos based scheme, the method not only provides with an improved scrambled performance, but also is easy to be carried out.(3) A blind approach based on sub-image histogram correlation is presented to objectively and automatically evaluate the degree of digital image scrambling. In the method, the definition and the feature of the ideal scrambled image is analyzed first, whose histogram is summarized at the same time. And then, the scrambled image is divided into some sub-images to construct some histogram sequences. Subsequently, image scrambling degree is assessed via the correlation of sub-image histograms. Finally, two kinds of experimental results indicate that the method not only is efficient, flexible, running without the origin image involved, but also can provide with some conclusions which are consistent with the perception of human visual system.(4) To evaluate the image scrambling degree effectively, another method of performance evaluation on image scrambling is designed via grey theory. The method first divides the scrambled image into several blocks, then computes the average of every line pixel values in the sub-images to static the gray information sequence of each block, and makes these sequences be small samples sequences. Finally the gray relevancy of every two sequences using gray relation analysis is calculated to evaluate the image scrambling degree. Experimental results show that the method can efficiently evaluate the scrambling effect of different scrambling transformation, running without the origin image involved. Compared with SNR, it can evaluate the scrambling effect more objectively and be consistent with the perception of human visual system.