Design And Implementation Of Virtual Restoration System For Damaged Cultural Relics

The classification and splicing of cultural relic fragments are the core of the virtual restoration process of damaged cultural relics,and are also the prerequisite for the data display of the Smart Museum.Terracotta Army fragments have the characteristics of small size,large similarity of individual fragments,lack of surface information,and severe wear of fracture surfaces,which results in low classification accuracy of fragments and poor splicing effect of fragments.In this paper,considering the problems in the process of fragment classification and fragment stitching,we propose a fragment classification and a stitching algorithm based on feature enhancement and multi-feature adaptive fusion,respectively.The main research contents include:(1)Fragment classification algorithm based on feature enhancement.In the process of cultural relic fragment classification,due to the lack of fragment feature information,the similarity of different fragment features is large,and the classification accuracy is poor,which indirectly leads to poor fragment mosaic effect.To tackle the above problems,the method applied in this paper is: firstly,feature enhancement processing is performed on the fragments.Specifically,considering the similar properties of different point features in the local space,the center point neighborhood spatial features are extracted by the multi-layer perceptron,and the extracted neighborhood space features are regarded as the feature increment.Hence,the feature enhancement of the fragments can be achieved.Secondly,the Point Conv convolution structure is combined with the Alex Net network model,to complete efficient feature extraction and classification on the 3D point cloud model of cultural relic fragments.Experiments show that the classification accuracy has been improved in the fragment classification task,using the fragment classification algorithm based on feature enhancement.(2)Fragment stitching algorithm based on multi-feature adaptive fusion.In the process of splicing cultural relics,it is difficult to obtain a good splicing effect by using a single feature to match,and it will also result in the high time complexity of the matching process using multiple features.In addition,the loss of cultural relics feature information is posing challenges to matched feature selection.Considering the above problems,our proposed method is to extract features such as curvature,torsion,left and right chord lengths from the fracture surface contours of cultural relic fragments,and then applies multi-layer perceptrons to perform feature fusion and feature compression on the feature vectors.The results of the multilayer perceptron will be used to splice the fragments.Finally,the error of fragment splicing is adopted to adjust the weight of the multilayer perceptron.Experiments show that the speed and effect of mosaic have been both improved in the fragment mosaic task,using the fragment mosaic algorithm based on multi-feature adaptive fusion.(3)Implemented a virtual restoration system for damaged cultural relics.The system mainly contains three modules,including import and export of cultural relics data,feature enhancement and classification of cultural relics data,and interactive splicing of cultural relics fragments,respectively.It is a relatively completed computer-aided cultural relics restoration system.