首页|基于小平面区域特征的文物碎片重组算法

基于小平面区域特征的文物碎片重组算法

扫码查看
针对非薄壁文物碎片重组复原过程中,因碎片边缘或断裂面磨损而导致拼接误差大的问题,提出了一种基于小平面区域特征的文物碎片拼接算法.首先根据区域增长分割把碎片的三维点云分割出许多小平面,建立以小平面作为拼接的特征区域,根据特征描述找到初始相似的小平面区域对,然后灵活应用几何约束+随机采样一致性(RANSAC)约束策略的方法,消除误匹配;最后通过 4PCS(4-Points Congruent Sets)算法进行碎片的粗对齐,再根据迭代最近点算法(ICP)实现碎片的精确拼接.实验结果表明:小平面的特征描述简单,且有较强的稳健性,采用两级约束策略的方法较好的解决了误匹配问题,提高了磨损碎片拼接的成功率,同时有效减少渗透现象.
Recombination algorithm of cultural relics fragments based on feature of small plane region
In the process of reorganization and restoration of non-thin-walled cultural relic fragments,the splicing error is large due to the wear of the edge or fracture surface of the fragments.A splicing al-gorithm of cultural relic fragments based on the feature of facet region is proposed.First,the 3D point cloud of the fragment was divided into many facets according to the region growth segmentation to es-tablish the feature regions with the facets as splicing.The initial similar facet region pairs were found according to the feature description,and then geometric constraints + random sampling consistency(RANSAC)constraint strategy method were flexibly applied to eliminate mismatches;finally,the 4PCS(4-Points Congruent Sets)algorithm is used to align the fragments roughly,and then the accu-rate splicing of the fragments is achieved according to the iterative closest point algorithm(ICP).The experimental results show that the feature description of the facet is simple and has strong robustness.The two-level constraint strategy method can better solve the mismatch problem,improve the success rate of wear debris splicing,and effectively reduce the penetration phenomenon.

non-thin-walled debrissmall plane regioncharacter descriptionrandom sampling consis-tencyiterative closest point algorithm

邱辉、谢晓尧、刘建成、王冲、刘嵩

展开 >

陆军勤务学院,重庆 401331

贵州师范大学 贵州省信息与计算科学重点实验室,贵州 贵阳 550000

四川省文物考古研究院,四川 成都 610041

非薄壁碎片 小平面区域 特征描述 随机采样一致性 迭代最近点算法

四川省重点研发计划四川省重点研发计划

2021YFS04012022YFS0558

2024

贵州师范大学学报(自然科学版)
贵州师范大学

贵州师范大学学报(自然科学版)

CSTPCD北大核心
影响因子:0.41
ISSN:1004-5570
年,卷(期):2024.42(1)
  • 5