Research On Matching Technology Of Cultural Relics Based On Salient Weight Feature Points And Multi-feature Constraints

The cultural relics have been buried deeply under the ground for thousands of years.When they were unearthed,the amount of debris was large and damaged to varying degrees,which brings great difficulty to the restoration of cultural relics.Digital restoration technology of cultural relics is a hotspot in the field of digital protection of cultural heritage.The existing virtual restoration methods mainly complete the splicing of fragments by extracting various features of the fracture surface.Due to the different degrees of damage to the fracture surface of the fragments,a single fracture surface matching method has been unable to achieve a good matching effect,and there are problems such as complicated matching conditions and low matching accuracy.In response to these problems,this paper proposes two matching methods according to the different fracture surfaces of cultural relic fragments.The content of this article is as follows:(1)For the cultural relic fragments with complete fracture surface information,the traditional surface feature point pair matching method has a complicated matching situation and pseudo matching between multiple sets of point pairs.A fracture surface matching method based on significant weight feature points is proposed.Firstly,the initial feature point set is constructed with the integral invariants of the points on the surface,and the weights of the feature points are defined.Based on the properties of the Huffman tree,the global similarity measurement of the fragmented fracture surface is performed to select some overall similar fragments;the global similarity measurement is carried out on the broken surface of the fragment,and some overall similar fragments are selected;then based on the Huffman tree,a significant weighted sequence set of feature points is constructed,and the potential matching sequence set is selected by judging the local region similarity of the feature points;finally,the matching is completed by combining the distance consistency and the concave-convex complementarity of the feature points.The experimental results show that this method can obtain relatively good matching results for the cultural relic fragments with less defects on the fracture surface.(2)For the fragments with serious missing geometric features on the fracture surface,multiple features on the fracture contour line are used for matching.In the traditional multi-feature fusion matching method,the correlation between each feature is not fully considered,and the weight contribution value cannot be effectively assigned to each feature,resulting in problems such as low matching accuracy.In this paper,a method of fracture contour matching based on feature metric constraints is proposed.Firstly,the point cloud model is meshed,and the edge contour line of the fracture surface is extracted,a multi-feature vector of the contour feature point set is constructed,and a factorization machine model is introduced.The model represents the constraint relationship between each feature matching metric item and solves the matching constraint function;Then a multi-fragment matching process of two-way search matching is proposed,which can solve the partial and complete matching problems of multiple fragments and improve the matching efficiency;Finally,through the rigid body transformation and the Iterative Closest Point algorithm to complete the splicing of fragments.Experiments show that this method can improve the matching accuracy and reduce the matching error.