三维点云配准算法的研究

Research on registration algorithm of three-dimensional point cloud

本文针对现有配准算法精度较低且适用范围有限的问题,提出了基于几何特征和RANSAC思想的粗配准算法以及基于点的邻域几何特征的迭代配准算法。该配准算法依据点的邻域曲率值提取两个点云的关键点集。配准过程中采用RANSAC算法的思想,每次采样中,利用FPFH特征来搜索对应点,并结合刚体变换不变量进一步约束,提高对应关系的准确性。经多次采样后,利用两点云一致性程度来选择最优的变换作为最终的变换关系。精配准过程依据最近点搜索法和点的几何特征初步确定候选点对,并结合本文提出的5维描述子和刚体变换不变量剔除错误的点对,提高对应关系的准确性,加快算法的收敛速度。

Aimed at the problem that the accuracy of the existing registration algorithm is low and its application range is limited,we present an initial registration method based on geometrical features and RANSAC algorithm, a precise registration method based on the geometric features of local neighborhood and the iterative thought. The initial registration algorithm extracts key points of two point clouds based on the curvature value of local neighborhood. The process of registration is based on RANSAC algorithm, with each sample uses FPFH to search corresponding points, and improves the accuracy of the correspondence relation according to the invariant constraints in rigid transformation. After multiple samples, this algorithm chooses the optimal transformation relying on the degree of consistency of two point clouds. The precise registration algorithm search initial corresponding points according to geometric features and the nearest point search method, and combines 5- dimensional description and invariant constraints in the rigid transformation to remove the wrong points in correspondence relation, which can improve the accuracy of correspondence relation of two point clouds and speed the rate of convergence.