基于PointNet++的碎片断裂面提取算法
Fragment Fracture Surface Extraction Algorithm Based on PointNet++
刘旭斌 1刘嵩 1刘建成 2王冲2
作者信息
- 1. 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 550001
- 2. 四川省文物考古研究院,四川成都 610041
- 折叠
摘要
针对文化遗产修复问题,提出了一种基于深度学习的碎片断裂面提取的算法,通过基于PointNet++的神经网络结构对碎片的断裂面进行了提取,并在FragTag3D数据集与3DPuzzle数据集上进行了测试.结果表明,该方法能够提取这些数据集中大部分碎片的断裂面.
Abstract
A deep learning based algorithm for extracting fragmented fracture surfaces was proposed to address the issue of cultural heritage restoration.The fracture surfaces of fragments were extracted using a neural network structure based on PointNet++,and tested on the FragTag3D dataset and 3DPuzzle dataset.The results indicate that this method can extract the fracture surfaces of most fragments in these datasets.
关键词
深度学习/PointNet++/语义分割/碎片拼接/断裂面提取Key words
deep learning/PointNet++/semantic segmentation/fragment reassemble/fracture surface extraction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据