改进TrICP算法的多源点云配准研究

Multi-Source Point Cloud Registration of Improved TrICP Algorithm

针对建筑物多源点云数据配准时出现配准精度低和效率不高的问题,提出一种基于主成分分析(PCA)的改进修剪迭代最近点(TrICP)多源数据配准方法。首先采用PCA算法对点云计算主轴方向,并基于方向反向问题创建校正矩阵进行校正,从而得到两组点云的初始良好位姿转换,完成点云的粗配准步骤;然后针对修剪迭代最近点中搜索点对和迭代次数问题,采用赋权的改进修剪迭代最近点算法完成两种点云数据的精配准。实验数据结果表明,改进的配准算法可以有效地减少多源点云数据迭代次数,提高配准精度并且保证了点云的完整性,相较于其他三种对比算法,精度分别提高56.75%、39.60%、28.08%,有效提升了配准的效率。

To address challenges related to low registration accuracy and efficiency in multi-source point cloud data registration for buildings,an enhanced Trimmed Iterative Closest Point(TrICP)method incorporating principal component analysis(PCA)is introduced.Firstly,the PCA algorithm is used to calculate the principal axis direction of the point clouds and create a correction matrix based on the direction reversal problem,so as to obtain the initial good positional transformation of the two groups of point clouds and complete the coarse alignment step of the point clouds;then,the improved Trimmed Iterative Closest Point algorithm is used to complete the fine alignment of the two kinds of point clouds for the problems of searching point pairs and iteration number in Trimmed Iterative Closest Point.The experimental data show that the improved alignment algorithm significantly reduces iteration counts,enhances alignment accuracy,preserves point cloud integrity,and boosts alignment efficiency.Comparative analysis shows accuracy improvements of 56.75%,39.6%,and 28.08%over three other algorithms,highlighting the effectiveness of the proposed approach.