Research On Fragment Restoration Technology Based On Surface Geometric Texture Features

In the process of repairing and restoring of a large number of damaged ceramic non-thin wall artifacts,due to its long history,after the natural weathering,man-made excavation and transportation,the debris of the cultural relics is vulnerable to be damaged,which is difficult to ensure the integrity of the fractured site.Most of them are damaged seriously.The traditional approaches based on the fracture contour and surface as matching features are not suitable for the situation above.And the effect is not necessarily significant,which has certain limitations.Therefore,in view of the serious defect of the fracture surface,the problem of large-scale restriction based on single feature matching is studied.This paper studies the virtual restoration technology of clay-based artifact debris with geometric texture features on the surface.The main research work of this paper are as follows:(1)Select the surface geometry texture as the main matching feature,and propose two texture feature shape definition methods.Aiming at the problem that the traditional threedimensional broken object restoration method has poor recovery effect on severe debris,the surface geometric texture is selected as the main feature,and the adjacent matching problem of the debris is transformed into the complementary problem between the surface texture shapes,which can reduce the dependence of geometric information of the fracture surface.The geometric texture features of the surface of the debris are extracted,and the texture shape information is calculated and determined.Two definitions of the texture element,based on the texture integrity and VE tuple,are proposed.(2)A matching method based on texture element integrity and chord length of the fracture profile is proposed.The geometric texture information of the surface of the broken artifact is used as the preliminary matching feature.According to the definition of the integrity-based texture element,the adjacent fragment matching constraint rule is constructed by the complementary relationship between the elements,and the constraint weight factor is added to the potential matching pair to avoid the time consumption and iterative redundancy brought by random selection.Combined with the boundary of the fracture boundary to execute the second constraint.The set of contour corner points is obtained,and the adjacent two points are connected into a string to form a chord sequence descriptor.The matching degree of the adjacent fragment contour is calculated by the longest common subsequence method.Combining the surface texture with the contour chord length for multi-feature matching avoids the limitation of a single feature and achieves a relatively reasonable and reliable matching result.(3)A matching method based on VE tuple and contour point set genetic algorithm is proposed.According to the definition of texture shape based on VE tuple,the matching relationship of VE tuple group is constructed.For the discrete contour points on the contour of the fracture contour,a set of discrete points is constructed,and the main features of the boundary contour are preserved in the case that the artifact entity is already defective.The gene chromosome descriptors are constructed by using the boundary point set,and the matching degree is calculated by the genetic algorithm fitness function.The experimental results show that the texture-based matching results is enhanced by the matching of the fracture contour boundaries.