Research On The Relic Fragment Matching Algorithm Based On Intuitionistic Fuzzy And Point Cloud Data

Cultural relics are the remains created by ancient people in the process of social development.These remains lay the foundation for the contemporary research on life customs,social conditions,technological level of ancient people,and are also a strong basis for modern restoration of historical truth.However,cultural relics are influenced by natural and man-made factors,and many precious heritage of history are damaged to varying degrees with the change of environment and times.Then computer aided virtual restoration of cultural relics can dramatically shorten the time of restoration and avoid secondary damage to cultural relics,which becomes an important research direction in computer graphics.However,there are still some problems and challenges of virtual restoration method in the practical application because of the serious damage of some cultural relics and the lack of significant characteristics.For example,general neural network can't effectively deal with the fuzziness of characteristics of cultural relics,which causes the poor classification performance of fragment of cultural relics.Meanwhile,there are some mismatches,the causes of which are that the multi-feature fitting algorithm of fragments doesn't consider the effect on damage of cultural relics.In view of the above problems,the main research contents of the thesis are as follows:(1)For the problem of low accuracy of PointNet classification caused by defect of cultural relic.In the paper,the theory of intuitionistic fuzzy and PointNet are combined,so as to propose a fragment classification algorithm of cultural relic based on PointNet.According to the theory of intuitionistic fuzzy,the fuzzified and defuzzified layer are built for PointNet,which reduces the effect of the fuzziness of the original data.Meanwhile,multi-scale spatial sphere is used to extract local features of point-cloud data,which reduces the effect of local feature missing in Poin Net on classification performance.(2)For the mismatches caused by the defect of cultural relics,the multi-feature matching model is proposed based on intuitionistic fuzzy and mixed optimization.Then the discriminant matrix is built on the basis of the historical fragment data,and the ternary method is used to blur the contour,thickness,curvature,binary descriptor of rotary projection of the fragment,so as to build the feature matric according to the fuzzification.Recognition model of the intuitionistic fuzzy and multi-feature is built based on the weighted similarity measurement of intuitionistic fuzzy and differential evolution algorithm.Finally,the recognition of the fracture surface on fragment of the cultural relics is completed.(3)For the problem of controlling cumbersome parameters setting of differential evolution algorithm in intuitionistic fuzzy multi-feature matching model,an adaptive differential evolution algorithm of parameter is proposed based on learning potential.Based on the Baldwin effect,the strategy of tapping individual learning potential is designed,and the adaptive transformation of crossed and mutagenic factors is built through individual learning potential,so as to improve the limitations of fixed parameters.Meanwhile,the effect of time factor on individual is considered,and the integration of time factor and learning potential improves the local search ability on lifting algorithm of selection operation...