在文物数字化的过程中，需要使用各种扫描设备来获取文物的形状信息。绝大多数扫描设备都需要经过多次扫描得到一组三维网格数据来重建文物模型。但由于设备精准度和操作问题，多次扫描得到的三维网格数据并非完全匹配，需要进一步对多个面片进行配准。 三维模型重建过程大致分为3大步骤:1)初步配准;2)精细匹配;3)面片融合，其中初步配准工作是所有工作的关键，也是三个步骤中至今未能满意解决的难题，这个问题在模型重构、模型识别和模型搜索领域举足轻重，也是这些研究领域的一大挑战。 本文主要针对三维网格数据初步配准问题展开研究，通过提取特征点、匹配特征点和空间变化矩阵求解三个步骤讨论了初步配准的关键问题，提出了一种基于曲率极大特征点匹配的三维网格数据初步配准算法。

Rigid 3D surface correspondence is an important problem in computer vision,used in tasks such as 3D modeling and 3D object recognition. Such problem occurs in using 3D-scanner.Most time,we can’t get the whole model of a object by once scanning,usually needs several times’.Due to the accuracy of equipments and operating, sometimes meshes are not matched well as expected. This thesis addresses “Rigid 3D surface correspondences problem”to resolve these problems. We present three stages to solve this problem: 1) coarse initial correspondence between surfaces; 2) compute accurate final correspondence between surfaces; 3) amalgamation of meshes. In this thesis, we focus on the first stage, and give an implement of some primary algorithm in it. First, we introduce the factual problem in model reconstruction using 3D-scanners. Then, some algorithms in getting feature-points are discussed in the sｅcond part.In the third part,method to find the relationship between two coordinate systems is the main topic. Introduction about calculating coarse initial correspondence is expatiated in the fourth part.Then, results are display in the fifth part. And, I summarize this thesis finally in the sixth part.