Research On Fragments Matching Of Damaged Cultural Relics Based On Multi-Feature

Cultural relics are generally broken and defective in the process of retention.As the number of fragments is large and the shape are different,artificial restoration is a long-term and difficult work.Using computer technology to realize virtual stitching of cultural relic fragments has become a hotspot in computer graphics.Due to the existing cultural relic fragment matching methods have high requirements on the integrity of geometric features,and have some problems such as single features,large errors and low efficiency,the matching effect is unsatisfactory.To overcome this problem,in this thesis,fracture surface features,thickness information,contour lines and other features are applied to matching and splicing of cultural relics fragments,and different matching methods are adopted to improve the matching effect according to the damage degree of the fracture site.The main research works of this thesis are summarized as follows:(1)For the fragments with complete fracture surface,aiming at the low accuracy of traditional matching method with single feature,a fragment matching method based on the adaptive neighborhood is proposed.Firstly,the salient feature points of the fracture surface are extracted,and the initial matching of the fragments of cultural relics is completed according to the similarity of the feature points.Then the adaptive neighborhood information is introduced as the upper constraint and the covariance matrix feature descriptor which combines the neighborhood concavity,convexity,curvature,and surface orientation is constructed.Finally,the similarity of the adaptive neighborhood of the initial matching point is judged,the pseudo matching point pairs are eliminated,and the optimal matching point set is selected by the spatial consistency test.The experimental results show that the method can obtain satisfactory matching results for the fragments with less defect on the fracture surface.(2)For the cultural relic fragments with serious fracture surface damage,aiming at the problem of large error in the traditional method of fracture surface matching,a cultural relic fragment matching method based on thickness and fracture contour features is proposed.In this method,the fragment thickness feature and the geometric features of fracture contour are combined,and multi-layer constraints is used to judge the matching relationship.Firstly,the thickness features are calculated and the thickness histogram is constructed,initial matching of fragments is realized according to the similarity of histogram Then the contour curve is segmented the geometric feature series of the sub-contour curve constructed.and an improved Hausdorff distance is introduced to judge whether there is a matched sub-contour curve in the initial matched fragment model.The experimental results show that this method has certain advantages over the traditional matching method for the cultural relic fragments with large damage on the fracture surface.(4)The virtual restoration system of cultural relic fragments is designed and implemented,which has the functions of model import,pre-processing,virtual matching and splicing of fragments,and export.The validity of the two fragment matching methods in this thesis is verified.The system is applied to the virtual restoration process of terracotta warriors and horses,and provides an auxiliary means for the entity assembly of terracotta warriors and horses fragments.