Depth Development Of Information With Graph And Image

Graphs and images are two kinds of important information complex carriers in the information age.In this paper,the method of mathematical calculation is applied to the frontier disciplines related to information,and the methods of information processing in complex carriers are deeply explored and developed.In the aspect of image depth mining,CoreRank algorithm is developed.By simulating the process of information transmission in the crowd,the influence of nodes in the network is calculated and the key nodes and core groups in the network are tapped.By deducing,we prove that the CoreRank algorithm iteratively converges on the inverse unit vector of one(or more independent and unrelated)full rank sub-matrixes.The result of the convergence depends on the set initial value and can be used to mine the central sub-Figure,that is the core(Core).The numerical experiments show that the proposed algorithm not only has low computational complexity and small size,but also can converge rapidly.At the same time,the computation results do not depend on the degree of nodes,and can find behind-the-scenes key nodes that other algorithms can not find.The CoreRank algorithm can also be used for search and organization mining of specific people.In the project of MLM network mining,we successfully tapped a number of organizational networks with the nature of MLM through the network,and achieved good results.It can be seen that CoreRank algorithm has a wide range of application prospects and research potential in key node discovery,core group search,and specific organization mining.In image information self-organizing,we focus on the classical image scrambling Arnold transform is only applicable to the square image defects,improved ArnoldX transform is designed for any parameter and matrix of arbitrary size,the algorithm does not increase the matrix size,not Changing the properties of the image itself,the calculation is small,the parameters do not depend on the matrix size,make up the defects of the traditional Arnold transform algorithm and other improved Arnold transform.At the same time,the ArnoldX transform retains the reversibility and periodicity of the Arnold transform,broadens the applicability of the Arnold transform and has long-term application prospect.In the aspect of image information security,we developed a kind of parallel encryption algorithm with high security,the r-type image encryption technology,which combines substitution,scramble and diffusion operations organically and scrambles globally by two-dimensional diffusion and ArnoldX algorithm.Encrypted ciphertext and plaintext have a strong nonlinear relationship.Numerical experiments show that the key and the encrypted intermediate data have a good combination of better self-adaptability,making the key and ciphertext satisfy the nonlinear requirements.Different from the current chaos-based encryption algorithm,we use a n-based random number table to generate pseudo-random sequences with high randomness and low computational complexity.In theory,the key space period can reach infinite.The algorithm also has parallelism and is suitable for the current GPU parallel environment,which can further improve the computational efficiency.Finally,we study the method of image generation with information features,encode the information into the random texture in the form of constraints,and design a variety of pseudo-natural texture generation methods widely used in industry.Muddy soil was generated by Markov random field method,pseudo-natural textures such as rain,bark and hemp rock were introduced by dispersion-adhesion model.Crack textures were designed to generate ice texture patterns.At the same time developed a variety of methods to solve the problem of seamless texture splicing.The research was selected as 2016 Development and Reform Commission Cocoon Fund Project.