Abstract: In order to avoid the drawback of the traditional ICP (iteration closest point) method for co-registration, a non-linear least squares co-registration model is proposed, which is based on minimizing the sum of squares of the Euclidean distances between two overlapped surfaces. The experiments demonstrate that the proposed method can be used for the co-registration of cloud points collected by the airborne laser scanner, and the experimental results are comparable with those of traditional methods. The experimental results imply that the proposed method has a good performance for transformation parameters estimation in 3D point clouds co-registration procedure. Especially, the non-linear co-registration method can achieve the best estimated value with only 6 to 7 iteration time. It is much faster than traditional method in practice, and can be treated as an extension for traditional co-registration theory of iteration closest point in the field of computer vision.

Key words: ICP, 3D surface matching, 3D similarity transformation, triangulated irregular network, conjugate points

摘要： 针对传统离散点云ICP配准算法迭代次数多、收敛慢等特点, 提出以3D点集间共轭点欧氏距离之和最小为理论基础的非线性最小二乘配准数学模型。通过对真实激光雷达扫描点云进行配准对比实验, 表明在初值条件相同情况下, 所提出的非线性ICP配准模型不仅能与传统ICP配准达到同样的配准精度, 而且能在6~7次左右达到快速迭代收敛, 是实现传统ICP配准模型的一种新途径。

关键词: 最近迭代点, 3D表面匹配, 空间相似变换, 不规则三角网, 共轭点