结合点云占比和平滑度的碗状碎块内表面识别算法
2020-08-26孙进丁煜王宁习俊通朱兴龙
孙进 丁煜 王宁 习俊通 朱兴龙



摘 要:基于内表面特性的碗状碎块重组算法可有效避免由断裂曲面引起的過分割问题,但其难点在于内表面的准确提取,为此提出一种结合点云占比和平滑度的碗状碎块内表面识别算法。首先基于区域生长算法将碗状碎块的表面分割成断裂曲面、底面、原始表面、内表面和外表面;然后根据其点云占比的显著差异从中识别出内表面和外表面,最后利用平滑度提取内表面。实验结果表明该算法能实现碗状碎块内表面的准确提取,并且具有更快的运算速度。
关键词:点云占比;平滑度;碗状碎块;内表面;识别
DOI:10.15938/j.jhust.2020.03.024
中图分类号: TP301.6
文献标志码: A
文章编号: 1007-2683(2020)03-0157-06
Abstract:The bowl-shaped broken pieces reassembly algorithm based on the inner surface characteristics can effectively avoid the over segmentation problem caused by the fracture surface, while the difficulty lies in the accurate extraction of inner surface. Therefore, an identification algorithm for inner surface of bowl-shaped broken pieces based on point cloud proportion and smoothness is proposed. Firstly, the surface of bowl-shaped broken pieces is divided into fracture surface, bottom surface, original surface, inner surface and outer surface based on the region growing algorithm. And then, according to the significant difference of point cloud proportion, the inner surface and outer surface are identified from the segmented surface group. Finally, the inner surface is extracted by the smoothness value. The experimental results show that this algorithm can accurately extract the inner surface of bowl-shaped broken pieces and has a higher calculation speed than other algorithm.
Keywords:point cloud proportion; smoothness; bowl-shaped broken pieces; inner surface; identification
0 引 言
人类悠久的文明使得河流、湖泊、海洋下隐藏着众多的文化遗迹。以近期结束清理工作的“南海一号”为例,从这艘沉船打捞出的十四多万件文物中有九成以上是以碗状器型存在的瓷器[1]。这些水下文物在水底经过长时间的侵蚀和颠簸,大多数变得破碎残缺。由于破损文物数量众多,利用计算机辅助技术将其数字化后进行虚拟拼接,对水下文物的保护有着十分重要的意义。为便于文物数字化表征,基于厚度值大致将破损物体分为两类:厚度小于1mm的如纸币、油画[2]和地图[3]等,称为碎片;厚度大于1mm的如碗、壁画[4]和兵马俑[5]等,称为碎块。考虑到水下文物中大部分是以碗状器型存在的物件,故本文以碗状碎块为研究对象。
在三维碎块拼接领域,传统的碎块重组算法主要围绕断裂曲面的几何特征[6]展开,因此许多专家学者提出了关于断裂曲面的识别算法。……
