Fragment Matching Of Terracotta Warriors Based On Their Thickness Feature

The Terracotta Warriors is an outstanding representative of ancient Chinese culture.However,due to various natural disasters and man-made destruction,it has been damaged when excavated.Archeologists repair artificially,which has the problems of low efficiency,high cost and difficulty in correcting errors.In recent years,the method of computer assisted virtual matching and splicing has become the mainstream,but it is affected by severe fragment damage and lack of details.This paper takes the fragments from the third excavated of pit one of terracotta warriors as experimental objects to research the fragment matching based on the thickness feature and solves the problem of the virtual reassembling of fragments when details are missing.The main research progresses of this thesis are summarized as follows:(1)A method for extracting the thickness feature of fragments is proposed,aiming at the required manual measurement in existing thickness-based matching methods.Firstly,divide the surface of the fragment model and calculate the set of contour curve points of the fracture surface from it.Secondly,the set of contour curve points is projected to a 2D plane with the maximum projected area.Finally,the thickness feature is obtained by calculating the distance of the top and bottom of the fracture surface contour line.The edge weight is calculated by the number of access to the vertices of adjacent triangles to improve the applicability of the contour extraction method based on edge weights.The rotation iteration method is used to reduce the number of projections.The contour lines are divided by relative direction and the parts on the upper and lower surfaces of the models are identified.Experiments have shown that this method can accurately extract the thickness feature of the fragment.(2)A method for fragment grouping based on thickness feature is proposed,aiming at a large number of useless matching in existing fragment matching methods.In this method,a thickness histogram is calculated using the thickness feature,and all the fragments are mapped into different groups by a Locality-sensitive hashing for Euclidean distance.The fragments in one group have similar thickness.The next searches for matching pairs are applied for the fragments in the same group,effectively reducing useless matching.(3)A rough matching method based on thickness feature is proposed.The thickness features are treated as an integer sequence,and the matching problem is converted into the longest common subsequence problem for each group.An improved method is used to solve the problem,and the matching result is obtained and roughly aligned.Experiments have shown that this method significantly reduces the time to find the matching pairs.(4)Designed and implemented a prototype system of the terracotta warriors fragment reassembling.Practice has proved that this system can assist in the virtual reassembling of fragment and help the physical repair of the terracotta fragments.