基于大尺度的点云精确配准算法

A Precise Registration Algorithm of Point Cloud Based on Large Scale

针对传统点云配准算法精度与尺度不能兼容的问题,本文提出了一种基于大尺度的点云精确配准方法.首先通过Kinect设备采集不同视角下目标物体的点云数据,然后通过对点云数据进行刚体变换实现各视角下点云数据坐标系的同一化,并对待配准的点云数据组进行尺度上的大幅度逼近,最后使用迭代最近点算法实现点云数据的精确配准.实验结果表明,该方法克服了迭代最近点算法仅适用于点云数据对距离较近的限制,且对迭代最近点算法的精确性没有影响,实现了在远距离的条件下对点云数据进行精确的配准.

Aiming at solving the problem of the incompatibility of the accuracy of the traditional point cloud registration algorithm with the scale, this paper presents a method of a precise registration of point cloud based on large scale. First, the point cloud data of the target object are collected with the Kinect de- vice, and then the same point of the cloud data coordinate system is obtained with rigid-body transtbrmation of the point cloud data. At the same time, we set a large approximation to the point cloud data, and finally we use of Iterative Closest Point algorithm to achieve the accurate registration of the cloud data. The experi- mental results show that the method can overcome the limitation in the point cloud data, while it does not af- tect the accuracy of the Iterative Closest Point algorithm, and it realizes the accuracy of the large-scale point cloud data registration.