激光点云粗配准方法实验对比与分析

Experimental Comparison and Analysis of Laser Point Cloud Coarse Registration Methods

点云粗配准要为点云精细配准提供良好的初始参数,其精度和效率是当今研究的热点问题。本文基于四组不同的激光点云数据,实现并对比分析了超级四点全等集算法(Super 4-Points Congruent Sets,Super4PCS)、四点全等集算法(4-Points Congruent Sets,4PCS)和基于软件Geomagic Studio 12的粗配准方法。实验结果表明,Super4PCS算法对于重叠率高,噪声大且特征明显的点云数据能够较好地配准;4PCS算法不适用于重叠率低的点云数据配准;软件Geomagic Studio 12适用于各种特点的点云模型,但该方法配准精度受人为因素的影响,配准效率低。

Coarse registration of point cloud needs good initial parameters for fine registration,its accuracy and efficiency are hot issues in current research.Based on four different sets of laser point cloud data,this paper realized and analyzed the super 4-points congruent sets(Super4PCS),the 4-points congruent sets(4PCS),and the coarse registration method based on software Geomagic Studio 12.The results showed that the Super4PCS algorithm could better register the point cloud data with high overlap rate,high noise and obvious characteristics;the 4PCS algorithm was not suitable for point cloud data registration with low overlap rate;the software Geomagic Studio 12 was suitable for point cloud models with various characteristics,but the registration accuracy of this method was affected by human factors and the registration efficiency was low.