Research On Virtual Splicing Of Damaged Figurines Based On Intuition-Fuzzy Swarm Intelligence Hybrid Strategy

The Terra-cotta Warriors and Horses in the Mausoleum of Emperor First Emperor of Qin were the golden signboards of the material heritage of the Chinese nation.However,due to the erosion of time and the accumulation of damaged cultural relics,high-performance computing is the necessary means to solve this problem.There are some problems,such as the irregular shape of terracotta fragments,fuzzy features,and difficulty of extract.And multi-fragment splicing is an NP problem.However,intuitionistic fuzziness has a stronger ability for representation and discrimination and is more suitable for multi-fragment matching with fuzzy features.In addition,swarm intelligence has an exclusive advantage in solving NP global stitching problems.Therefore,this paper combines intuitive fuzziness with a swarm intelligence algorithm and applies it to multi-fragment virtual splicing.The research is as follows:(1)To improve the diversity and convergence ability of the differential evolution algorithm,the parameter adaptive adjustment mechanism and improvement strategy that affect the performance of the differential evolution algorithm are analyzed and summarized.And an adaptive differential evolution algorithm based on elite differential feedback strategy(EDF-DE)is proposed.Firstly,the neighborhood model is established by diversity weighted distance index screening.Secondly,an elite difference feedback crossover strategy is designed by searching for global and neighborhood elite differences,and the two differences guide the crossover operator interactively.Finally,an adaptive method is used to dynamically control the step size factor,and a mutation operator with archiving strategy is introduced to balance the convergence and diversity of the DE algorithm.(2)In the fuzzy multi-objective optimization problem,the fuzzy set can only represent the limitation of a single membership degree,and thus cannot describe the distribution of the current solution set in detail.A gray wolf multi-objective optimization algorithm based on the intuitionistic fuzzy entropy hybrid strategy is proposed(MOIF-MSGWO).Firstly,the quality and diversity of the initial population were enhanced by quadratic interpolation.Secondly,intuitionistic fuzzy population entropy(IFE)is designed to establish three control strategies for IFE fine-grained distribution.Based on the degree of distribution and aggregation embodied in IFE.The grey Wolf update operator and self-learning operator are designed for local development,and the multiplication and division operator is used to achieve global optimization,obtain the uniform distribution of multi-objective solutions,and improve the accuracy of Pareto solutions.(3)The improved multi-target and single target are applied to matching and splicing the fragments of terracotta figures respectively.To solve the problem of feature loss and NP-hard matching,the MOIF-MSGWO algorithm was used to solve the optimal matching pairs of multi-fragments,and the mapping relationship between individuals in the Pareto solution set and the global matching table was established.The matching table is further used to get the optimal matching pair and eliminate the mismatched fragments.On this basis,the piecing task was modeled as a single-objective optimization problem,and the EDF-DE algorithm was used to solve the optimal individual iteratively.With individual parameters as rotation and translation indices,the transformation matrix was constructed to complete the global fine piecing of multiple fragments.