航空学报 > 2023, Vol. 44 Issue (1): 326336-326336   doi: 10.7527/S1000-6893.2021.26336

基于测地线距离的空间非合作目标点云配准

陈斌1(), 郗厚印1, 张晓东2, 罗敏2, 郭治栋1   

  1. 1.北京邮电大学 人工智能学院,北京 100876
    2.北京空间飞行器总体设计部,北京 100094
  • 收稿日期:2021-09-07 修回日期:2021-10-25 接受日期:2021-11-12 出版日期:2023-01-15 发布日期:2021-11-23
  • 通讯作者: 陈斌 E-mail:binchen@bupt.edu.cn
  • 基金资助:
    国家自然科学基金(62073033)

Point cloud registration of space non-cooperative targets based on geodesic distance

Bin CHEN1(), Houyin XI1, Xiaodong ZHANG2, Min LUO2, Zhidong GUO1   

  1. 1.College of Artificial Intelligence,Beijing University of Posts and Telecommunications,Beijing 100876,China
    2.Beijing Institute of Spacecraft System Engineering,Beijing 100094,China
  • Received:2021-09-07 Revised:2021-10-25 Accepted:2021-11-12 Online:2023-01-15 Published:2021-11-23
  • Contact: Bin CHEN E-mail:binchen@bupt.edu.cn
  • Supported by:
    National Natural Science Foundation of China(62073033)

摘要:

针对局部结构相似及翻滚遮挡下目标动态特性辨识问题,提出一种融合疏密度指标与全局测地线距离的空间非合作目标点云配准方法。首先,引入球面投影法,将不规则的空间目标点云映射到规则的球面流形上。然后,设计基于点云离散程度的疏密度评价指标,将球面点云划分为不同的局部点云子集,在此基础上构建全局测地线距离矩阵以增强点云局部形状信息的感知能力。最后,依据全局测地线距离矩阵推导建立了场景点云与模型点云间配准矩阵,实现空间非合作目标的动态特性辨识。仿真与试验结果表明:提出的算法在目标局部结构相似及点云缺失下的配准精度优于最近迭代点(ICP)算法与基于凸包粗配准的改进ICP算法。

关键词: 非合作目标, 点云配准, 球面投影, 测地线距离, 疏密度

Abstract:

To realize identification of the dynamic characteristics of the space targets with local structural similarity and tumbling occlusion, a novel point cloud registration method is proposed for the space non-cooperative target by using the density index and global geodesic distance. First, the spherical projection method is introduced to map the irregular point cloud of the space target to the regular spherical manifold. Then, a density evaluation index based on the dispersion degree of point cloud is designed to divide the spherical point cloud into local subsets. On this basis, a global geodesic distance matrix is constructed to enhance the perception of the information of the local shape of the point cloud. Finally, according to the global geodesic distance matrix, a registration matrix between the scene point cloud and the model point cloud is established to realize the dynamic characteristic identification of the space non-cooperative target. Simulation and experimental results show that with similar target local structures and missing point clouds, the proposed algorithm has better registration accuracy compared to Iterative Closet Point(ICP) algorithms and improved ICP algorithms based on convex hull coarse registration.

Key words: non-cooperative target, point cloud registration, spherical projection, geodesic distance, density index

中图分类号: