基于法向量和密度的点云特征点提取问题研究  

FEATURE POINTS EXTRACTION OF POINT CLOUD BASED ON NORMAL VECTOR AND DENSITY

作　　者：马聪聪[1] 李松[1] 曹菁菁[1] 于蒙[1] Ma Congcong;Li Song;Cao Jingjing;Yu Meng(School of Logistics Engineering,Wuhan University of Technology,Wuhan 430063,Hubei,China)

机构地区：[1]武汉理工大学物流工程学院,湖北武汉430063

出　　处：《计算机应用与软件》2020年第5期256-260,292,共6页Computer Applications and Software

基　　金：国家自然科学基金项目(61502360,71672137)。

摘　　要：在点云数据处理领域,针对现有特征点提取算法容易出现错判与漏判的问题,提出一种基于点云法向量与密度相结合的特征点提取算法。在点云模型中构建拓扑关系,采用KD树的方法建立目标点的K邻域;计算各点云数据对应的法向量并将邻域内各点的法向量点积的平均值作为特征度;以点间距离为依据对点云密度进行计算;将特征度与点云密度相结合组成特征点判别参数,与阈值进行比较实现特征点提取。实验结果验证了算法的可行性与有效性。In the field of point cloud data processing,aiming at the problem that the existing feature point extraction algorithms are prone to misjudgment and omission,this paper proposes a feature point extraction algorithm based on the combination of point cloud normal vector and density.The topological relationship was constructed in the point cloud model,and the K-neighborhood of the target point was established by using the KD tree method;we calculated the normal vector corresponding to each point cloud data and took the average value of the normal vector dot product of each point in the neighborhood as the feature degree;the point cloud density was calculated based on the distance between the points;the feature degree and point cloud density were combined to form the feature point discriminant parameters,and the threshold value was compared to realize the feature point extraction.The experimental results validate the feasibility and effectiveness of our algorithm.

关 键 词：点云 特征点提取 法向量 特征度 点云密度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]