基于轮廓线特征点的交互式文物拼接

发布时间：2018-06-06 19:33

本文选题：虚拟复原 + 交互式算法　；参考：《光学精密工程》2017年06期

【摘要】：针对计算机辅助文物虚拟复原中断裂部位受损而引起的几何特征丢失问题,提出一种结合表面纹饰信息与轮廓线上特征点的交互式破损文物拼接算法。首先,通过生成拉普拉斯线得到碎片表面的纹饰特征和断裂部位表面轮廓线。领域专家根据纹饰特征的连续性标记表面约束点,再采用Douglas-Peucker算法提取断裂部位轮廓线上的特征点;然后,结合特征点至约束点的距离、特征点与邻接点的内角以及相邻特征点的欧氏距离构造能有效表示断裂部位拼合关系的特征点描述符,同时定义匹配度函数从而获取特征点匹配对集合;最后,采用四元组方法计算刚体变换矩阵,实现碎片的精确拼合。实验结果表明,拼合时间节省了约13%~16%,拼合误差小于1mm。该方法避免了传统的断裂面拼合方法因数据量大带来的高时间代价,可快速有效地实现破损文物碎片模型的虚拟复原。
[Abstract]:In order to solve the problem of geometric feature loss caused by the damage of fracture site in computer-aided virtual restoration of cultural relics, an interactive splicing algorithm for damaged cultural relics is proposed, which combines the surface decoration information with the feature points on the contour line. Firstly, the decorative features of the fragment surface and the surface contours of the fracture site are obtained by generating Laplace lines. According to the continuous marking of the decorative features, the domain experts use the Douglas Peucker algorithm to extract the feature points on the contours of the fracture site, and then combine the distance between the feature points and the constraint points. The Euclidean distance between feature points and adjacent points and Euclidean distance of adjacent feature points can effectively represent the feature point descriptors of the splicing relationship of fault sites, and the matching degree function is also defined to obtain the set of matching pairs of feature points. A quaternion method is used to calculate the rigid body transformation matrix, and the exact splicing of fragments is realized. The experimental results show that the stitching time is saved about 13 / 16 and the splicing error is less than 1 mm. This method avoids the high time cost caused by the large amount of data, and can quickly and effectively restore the debris model of damaged cultural relic.
【作者单位】：北京师范大学信息科学与技术学院;西北大学信息科学与技术学院;
【基金】：国家自然科学基金面上项目(No.61673319,No.61373117) 高等学校博士学科点专项科研基金(No.20136101110019)
【分类号】：TP391.41

【相似文献】