电子学报 ›› 2022, Vol. 50 ›› Issue (6): 1436-1443.DOI: 10.12263/DZXB.20201236

• 学术论文 • 上一篇    下一篇

基于特征区域划分的文物碎片自动匹配算法

赵夫群1, 耿国华2   

  1. 1.西安财经大学信息学院,陕西 西安 710100
    2.西北大学信息科学与技术学院,陕西 西安 710127
  • 收稿日期:2020-11-02 修回日期:2021-02-28 出版日期:2022-06-25
    • 作者简介:
    • 赵夫群 女,1982年生,山东临沂人.博士,现为西安财经大学信息学院副教授.主要研究方向为图形图像处理,三维重建.E-mail: fuqunzhao@126.com
      耿国华 女,1955年生,山东莱西人.现为西北大学信息科学与技术学院教授,博士生导师.主要研究方向为计算机辅助文物数字化复原.E-mail: ghgeng@nwu.edu.cn
    • 基金资助:
    • 国家自然科学基金 (61731015); 陕西省自然科学基础研究计划 (2021JQ-765); 陕西省哲学社会科学重大理论与现实问题研究项目 (2021ND0141); 西安财经大学科研计划 (20FCJH002)

Automatic Matching Algorithm of Cultural Relic Fragments Based on Feature Region Division

ZHAO Fu-qun1, GENG Guo-hua2   

  1. 1.School of Information, Xi’an University of Finance and Economics, Xi’an, Shaanxi 710100, China
    2.School of Information Science and Technology, Northwest University, Xi’an, Shaanxi 710127, China
  • Received:2020-11-02 Revised:2021-02-28 Online:2022-06-25 Published:2022-06-25
    • Supported by:
    • National Natural Science Foundation of China (61731015); Natural Science Basic Research Program of Shaanxi Province (2021JQ-765); Philosophy and Social Science Major Theoretical and Practical Issues Research Program of Shaanxi Province (2021ND0141); Research Program of Xi 'an University of Finance and Economics (20FCJH002)

摘要:

在破损文物虚拟复原过程中,碎片匹配是重要研究内容之一,而传统几何驱动匹配算法对断裂面重叠率较低碎片的匹配误差较大,因此针对具有一定厚度的文物,本文提出一种基于特征区域划分的文物碎片自动匹配拼接算法.首先提取文物碎片的断裂面,并在断裂面上点的局部区域内构造法向量特征,计算断裂面的特征点和特征点集;然后将特征点集划分为多个小规模的点云区域,并通过区域匹配将碎片断裂面初步对齐;最后利用基于阈值约束的迭代最近点(Iterative Closest Point,ICP)算法对特征点进行精确对齐,从而实现碎片的最终匹配.实验结果表明,该基于特征区域划分的文物碎片匹配算法的时间效率和匹配精度,比多数已有算法至少可以分别提高10%和20%,适用于低重叠断裂面的文物碎片匹配.

关键词: 碎片匹配, 特征点, 区域划分, 阈值约束, 迭代最近点

Abstract:

In the process of virtual restoration of damaged cultural relics, fragments matching is one of the important research contents. The traditional geometric driven matching algorithms have large matching error for fracture surfaces with low overlapping rate,thus aiming at cultural relics with certain thickness, an automatic matching and splicing algorithm of cultural relics fragments based on feature region division is proposed. Firstly, the fracture surfaces of fragments are extracted, normal vector features are constructed in the local region of the points on the fracture surface to obtain the feature points and feature point sets of the fracture surfaces; secondly, each feature point set is divided into several small-scale point cloud regions, and the fragment fracture surfaces are aligned by region matching; finally, the iterative closest point (ICP) algorithm based on threshold constraint is used to further accurately align the fracture surfaces, so as to achieve the final matching of cultural relic fragments. The experimental results show that the time efficiency and matching accuracy of the proposed algorithm are at least 10% and 20% higher than those of many existing algorithms, respectively, and it is suitable for the matching of cultural relic fragments with low overlapping fracture surfaces.

Key words: fragments matching, feature point, region division, threshold constraint, iterative closest point

中图分类号: