Research On Virtual Matching Method Of Cultural Relic Fragments Based On Optimized Feature Constraints

Pottery cultural relics are an important historical witness of social civilization and development.Generally,pottery cultural relics are made of clay.Influenced by the natural environment and man-made destruction factors,most of them are fragmented and have vague surface characteristics when excavated.As a representative of ceramic relics,terracotta warriors are also faced with the problems of serious damage,large number of fragments and disorderly excavation,which brings great challenges to the restoration work.Because the existing methods of fragment matching and hole repair have high requirements for the integrity of geometric features,and have the disadvantages of a single feature,large error and low efficiency,the repair effect of damaged terracotta warriors is not ideal.Aiming at the problems of large amount of data,uneven distribution of data points and high noise content in the three-dimensional data model of terracotta warriors' fragments,this thesisr mainly studies the key technologies in the virtual restoration of cultural relics,such as the matching of fragments' fracture surfaces and hole restoration,and the specific research work is as follows:(1)A fragment matching method based on GH-LS3 D and simulated annealing ICP was proposedFirstly,the GH-LS3 D algorithm is used to rough match the fracture surface of the fragments;Then,the ICP algorithm based on scale factor and simulated annealing coefficient is used to achieve accurate matching of fracture surface,so as to achieve matching of cultural relics fragments.The results show that this method can effectively improve the matching accuracy of point cloud data model with large amount of data and large resolution difference,and realize the matching of cultural relics fragments.(2)A fragment matching method based on region segmentation is proposedFirstly,the concave and convex feature points on the fragment fracture surface are extracted;Then,the feature points are segmented by using the segmentation algorithm based on the mixed manifold spectrum clustering;Finally,the ICP algorithm based on SVD is used to match the region,so as to achieve accurate fragment matching.The matching method is applied to the public data sets and the Marsh cultural relics data sets.The results show that the proposed method can improve the overlap rate of regional point clouds and effectively solve the matching of low-overlapped artifacts.(3)A fragment matching method based on extended PFH feature was proposedFirstly,the EPFH feature descriptors with significant feature points on the fracture surface of the fragment are constructed.The descriptor has a strong discriminant degree;Then the sampling consistency algorithm is used to estimate the rigid body transformation of the fracture surface matching to achieve the initial matching.Finally,the ICP algorithm based on k-d tree is used to realize fine matching of the fracture surface.The matching method is applied to the cake common debris data sets and the military corpus artifact data sets.The results show that the proposed method can solve the splicing of cultural relics with high noise content.(4)A method of radial basis function for hole restoration of cultural relics based on feature line subdivision was proposed.This method uses the radial basis function to construct the implicit surface and subdivides the hole area through the feature line,which is a global repair method.In the hole repair process,the process of spatial segmentation and integration can be avoided,and the hole detail features are effectively captured,so that the repaired surface is closer to the original real shape and has a good smooth connection with the surrounding mesh.By applying the repair method to the public data sets of bunny and horse and the data sets of Terracotta Warriors,the results show that the method has good effect on the repair of holes with different types and different geometric features.