In this note, we propose an approach for initializing the Iterative Closest Point (ICP) algorithm that allows us to apply ICP to unlabelled point clouds that are related by rigid transformations. We also give bounds on the robustness of our approach to noise. Numerical experiments confirm our theoretical findings.
翻译:在本说明中,我们提出了一个初始化热点(ICP)算法的方法,使我们能够将比较方案应用于与僵硬变换有关的无标签点云层。我们还为我们对噪音的处理方法的稳健性提供了界限。数字实验证实了我们的理论结论。