Swarm Intelligence Algorithm And Its Application In Virtual Merging Of 3D Cultural Relics

It is widely approved that the Swarm Intelligence (SI) algorithms are effective for solving complex optimization problems in engineering fields. However, since the SI algorithms are originated from biological simulation and involve sophisticated stochastic swarm behaviors, development and application of the traditional SI algorithms are limited because of their weak convergence theoretical foundation and the problem of slow convergence speed and premature convergence. There exist poor recognition ability for adjacent fragments and poor robustness in virtual merging of 3D cultural relic’s fragments and their mathematical model are all complex optimization problems. This thesis focuses on the global convergence of swarm intelligence optimization algorithm for the construction of guaranteed global convergence algorithms, and the optimization problems of the virtual merging of 3D cultural relic’s fragments are solved by optimization method based on swarm intelligence optimization algorithms. The main contributions of this thesis include:(l)Propose two improved SCO algorithms to further improve the global search ability of SCO algorithmAiming at the problems of the lack of global convergence theoretical analysis for traditional social cognitive optimization (SCO) and exploration & exploitation ability needing to be further improved, the global convergence proof of the traditional SCO algorithm is given. Then a hybrid social cognitive optimization (HSCO) algorithm based on elitist strategy and chaotic optimization is proposed; a quantum-behaved social cognitive optimization (QSCO) algorithm based on quantum delta potential well model is proposed. Improved social cognitive optimizers are guaranteed to convergence to the global optimization solution, and have higher global convergence speed, which lay the theoretical foundation for expanding the application range of SCO algorithm.(2)Propose chaotic BCC optimization algorithm to improve the exploration ability of BCC optimizationAiming at the problems of the lack of theoretical analysis for bacterial colony chemo taxis (BCC) optimization algorithm and its slower convergence speed,the global convergence proofs of bacterial chemo taxis optimization algorithm (BC) and BCC optimization algorithm are given. A chaotic bacterial colony chemo taxis (CBCC) optimization algorithm based on chaotic optimization is proposed.Chaotic optimization with synchronous iteration and sharing the optimal search results in bacterial group and chaotic group is introduced to keep the variety of bacterial movement and enhance the ability of exploration and convergence speed.(3) Propose a recognition method of multi feature extracting and intelligent fusion to enhance recognition ability for adjacent fragmentsAiming at the problems of inaccurate feature information from 3D cultural relic’s models, a new integrated algorithm of extracting multi feature from cultural relic fragments is proposed, a multi feature matching recognition method with intelligent multi feature fusion is proposed based on evidential reasoning and interval number. The new method solves uncertain fuzzy matching problem in the single feature information, and enhances recognition ability for adjacent fragments, which lays the foundation for global virtual merging of 3D cultural relic’s fragments.(4) Propose a merging method of global optimal matching and intelligent registration on SI algorithms to enhance the robustness of merging methodAiming at the lack of overall consideration in global matching and the problem of registration without mutual information and accurate corresponding points of 3D fragments, Build mathematical models of global matching of fragments for two kinds of global matching problems, then use discrete CBCC algorithm to solve the problem.The optimal corresponding points in coarse registration are obtained by discrete CBCC (DCBCC) algorithm and the optimal coordinate transform is estimated by CBCC algorithm. The new method improves the global optimization ability and robustness of the matching algorithm, improves the precision and efficiency of registration, and expands the scope of application of the registration algorithm. To sum up, this thesis includes two aspects:SI algorithm convergence and SI algorithm application. The global convergence proof of the proposed SI algorithms is given in theory, which improves algorithm performance and expands its applicable domain. The key problems of the virtual merging of 3D cultural relic’s fragments are solved by optimization method based on SI algorithms, which promotes the application of SI algorithms in 3D model.