邻域密度约束的动态标准差阈值三维点云数据离群点检测方法  

Outliers Detection Method Based on Dynamic Standard Deviation Threshold Using Neighborhood Density Constraints for Three Dimensional Point Cloud

作　　者：杨永涛 张坤[3] 黄国言[2] 吴培良[2] Yang Yongtao;Zhang Kun;Huang Guoyan;Wu Peiliang(Information Technology Center of Yanshan University, Qinhuangdao 066004;School of Information Science and Engineering, Yanshan University, Qinhuangdao 066004;School of Information Science and Engineering, Hebei University of Science and Technology, Shijiazhuang 050018)

机构地区：[1]燕山大学信息技术中心,秦皇岛066004 [2]燕山大学信息科学与工程学院,秦皇岛066004 [3]河北科技大学信息科学与工程学院,石家庄050018

出　　处：《计算机辅助设计与图形学学报》2018年第6期1034-1045,共12页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61305113);河北省自然科学基金(2016203358);燕山大学博士基金(BL18002)

摘　　要：为了提升三维点云数据离群点的检测能力,提高检测方法的适应性,解决针对密度分布变化大的点云数据离群点检测效果不佳的问题,提出一种基于邻域密度约束的动态标准差阈值三维点云数据离群点检测方法.该方法充分考虑获取的点云数据的密度差异,将点云的密度特征引入离群点判定阈值的计算.首先利用直通滤波提取目标点云数据,检测并移除无效点;然后分析离群点的检测原理,给出点云k-邻域密度的估算方法;最后通过邻域密度约束实现了标准差阈值的动态调整,并采用不同的约束方式对远离主体点云的外部区域和内点区域的离群点进行检测,实现了密度分布变化明显的点云数据离群点的有效检测.实验结果表明,文中方法能够更加有效地移除离群点,通过标准差阈值动态约束满足了密度分布差异较大的点云数据的针对性检测,提升了检测效果和检测性能,达到了预期的目的,对实际应用具有积极意义.In 3 D point clouds reverse engineering, the outliers detection plays a key role on the subsequent processing. However, when the point cloud has big change density distribution, the detection of outliers becomes very difficult. In order to get a feasible detection result, improve the detection ability and adaptivity, an outliers detection method was proposed, which based on dynamic standard deviation threshold using k-neighborhood density constraints. This method fully considered the density difference of the obtained point cloud, and introduced the density characteristics into calculation of the determining threshold. Firstly, the target point cloud by pass-through filtering was extracted, and the invalid points were removed. Then the detection principle was analyzed, the k-neighborhood density estimation method was presented. Finally the dynamic standard deviation threshold constrained by the k-neighborhood density was calculated, the different constraints for outer regions and inlier regions were obtained, and a better detection result for point cloud with big change density distribution was got. Experimental results show that the method can apply to the highly variable density distribution point cloud, get a feasible detection result, improve detection effect and performance, and is positive to practical applications.

关 键 词：离群点检测 邻域密度 判定阈值 标准差 内点点云 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]