自定义聚类中心点的快速K-means聚类点云精简算法
Fast K-means Clustering Point Cloud Reduction Algorithm for Custom Clustering Center Points
王世刚 1关红利1
作者信息
- 1. 广西科技大学自动化学院,广西 柳州 545616
- 折叠
摘要
针对传统K-means算法在随机选取聚类中心点出现聚类失败及点云数据重建时在相对平坦的区域出现孔洞的问题,提出一种GK-means的改进聚类算法对点云数据进行精简.该算法首先规定数值K作为最终聚类个数的限定,然后对选点策略进行改进,采用最远点采样选取聚类中心,对簇进行细分,计算所有点到聚类中心的欧氏距离,获取最小值所在的位置,放进最小距离所在的簇.实验结果表明:改进后的K-means算法能够使算法成功的概率提高且运行速度较快,对点云进行精简时,特征区域完整地保留了点云模型的细节特征,重建结果具有较高的光顺性.
Abstract
An improved clustering algorithm of GK-means is proposed to simplify point cloud data in this paper.The al-gorithm first stipulates the value K as the limit of the final number of clusters,and then improves the point selection strate-gy,selects the cluster center with the farthest point sampling,subdivides the cluster,calculates the Euclidean distance from the point to the cluster center,obtains the position of the minimum value,and puts it into the cluster where the minimum distance is located.The experimental results show that the improved K-means algorithm can improve the success probabili-ty of the algorithm and run faster.When the point cloud is simplified,the feature region can retain the detailed features of the point cloud model completely,and the reconstructed results have high smoothness.
关键词
聚类中心/迭代/GK-means算法/点云精简Key words
cluster center/iteration/GK-means algorithm/point cloud reduction引用本文复制引用
出版年
2024
NETL
NSTL
万方数据