Image Analysis And Encryption

The analysis of digital image is an effective approach based on video technology and artificial intelligence. By the algorithm, we can make our life more automatical. Encryption techniques of digital images are very practical and crucial techniques. This thesis makes a depth research on these two fields, and its major contributions are as follows:1st. It introduces a totally new view to analysis the whole procedure of image analysis. By adopting the view, we can pick up the difference among different image analysis algorithm with more efficiency. We can also make the improvement of the existing algorithm clearer and take much more purposely steps to propose new image analysis algorithm.2nd. It introduces the definition of texture direction. Then, it proposes a sorting and partition algorithm of images with directional texture. Also, it discusses the origination of texture images and non-texture images, the essence of texture and issues to be noticed when dealing with textures. After that, the thesis proposes a algorithm to seek the degree of the texture of a digital image.3rd. It gives a new model for data clustering, "village-town" model and a clustering globally best algorithm based on genetic method. Taking data to be clustered as objects and beginning with finding globally best configuration. It is better than FCM in applicability and precision.4th. A theorem provides a new method for digital image encryption. And it has a good anti-decoding property against traditional techniques for decoding encrypted images.In this thesis, several methods such as wavelet analysis, fuzzy theory, data clustering, neural network, model recognition and genetic algorithm are mentioned. And the author has made some improvement in fields of digital image analysis and encryption with genetic algorithm and wavelet analysis. Besides, some theory such as rough set, data information and morphologic are also summarized by the author about their development situation and trend in image analysis and encryption in this thesis. Hope they are useful for readers.