摘要
针对激光线结构光扫描仪等获取的复杂物体表面三维点云数据量大、冗余性高等问题,提出一种基于自适应邻域和局部贡献值的点云精简算法。首先,根据点云的局部几何特征确定每个区域的最佳邻域范围;其次,将最佳邻域与内部形状特征算法、局部表面斑块算法相结合,计算所有点云数据的局部贡献值,并提取出点云特征点;最后,使用K-means聚类算法划分点云数据,并按类别和贡献值对点云进行精简。实验结果表明,对于复杂表面被测物,所提算法能够在保证精简率的情况下调节特征区域与非特征区域的精简度,同时保证点云完整性与细节特征信息,精简结果精度较高并更贴合物体原始面貌。
To address the problems associated with the large amounts of data and high redundancy of three-dimensional point cloud related to complex object surfaces obtained by a laser line structured light scanner,a point cloud simplification algorithm based on self-adaptive neighborhood and local contribution value is proposed.First,according to the local geometric characteristics of the point cloud,the best neighborhood range is selected.Then,the best neighborhood,internal shape feature algorithm,and local surface patch algorithm are combined to calculate the local contribution values of all point cloud data and the feature points of the point cloud are extracted.Finally,K-means clustering algorithm is used for classification and the point cloud is simplified based on the classification results and the contribution values.The experimental results show that for complex surface test objects,the proposed algorithm can adjust the simplification of characteristic and noncharacteristic areas while ensuring the simplification rate as well as the overall integrity and detailed feature information of the point cloud.Consequently,the simplification result has higher accuracy and fits the original appearance of the object more closely.
作者
郑茹丹
李金龙
张渝
高晓蓉
Zheng Rudan;Li Jinlong;Zhang Yu;Gao Xiaorong(School of Physical Science and Technology,South uvest Jiaotong University,Chengdu,Sichuan 611756,China)
出处
《激光与光电子学进展》
CSCD
北大核心
2021年第16期321-328,共8页
Laser & Optoelectronics Progress
基金
国家自然科学基金(61771409)。
关键词
图像处理
点云精简
自适应邻域
特征区域划分
K-MEANS
image processing
point cloud simplification
adaptive neighborhood
feature area division
K-means
