Research On Splicing Method Of Terracotta Fragments Based On The Matching Of Fracture Surface

With the rapid development of computer digitization technology, visualization technology and computer equipment, the excavation and protection of the cultural relics are also gradually in the digital stage. The restoration of the broken cultural relics is an important part for the Heritage protection. The computer-aided relic restoration can greatly shorten the cycle of artificial restoration, reduce the difficulty of manual work and avoid the second damage to the cultural relics So, this technology has great value and meaning for further research. In this paper, based on the broken pieces of Terracotta Army Mausoleum of the First Qin Emperor, we makes an in-depth study on the virtual mosaic technology of the non-wall cultural relics. The research of this paper is as follows:(1) An improved region growing algorithm for surface segmentation is proposed. In order to solve the problems of the poor robustness of curvature, we compute the integral invariants of terracotta pieces surface under multi-scale radius. Then we make classification on the feature points which is used for segmentation though the integral invariants. After that, according to the classified feature points, we set weights on triangular mesh. And then we fragment the surface segmentation based the weight-region growth algorithm and boundary removed region merging algorithm. Finally, we combine rough degree with area to identify the fracture surface. This method can effectively solve the problem of segmentation.(2) A fracture surface matching algorithm based feature region matching is presented. First, calculate the unevenness of the fracture surface vertexes and classify the surface vertexes. Then generate a feature region of the similar vertexes according to the connected clustering method. Calculate the similar feature region with the principal component analysis (PCA) method and obtain optimum feature region matching collection with coarse matching implementation through compatibility constraints and error measures.(3) On the basis of the optimum feature region matching collection, a VSVD-ICP algorithm is presented to further stitch debris precisely. This method introduces integral invariants constraints to SVD algorithm and introduces the unevenness of the feature points to the conventional ICP algorithm. Experiments show that the proposed method not only achieve precise stitching of the two adjacent Terracotta pieces, but also speed up the iterative guide of matching alignment and improve the accuracy of matching.(4) We design and realize the system of implements Terracotta debris splicing. The system is easy to operate, streamlined and practical. It works well on putting together Terracotta debris.