基于Super4PCS的激光点云配准适用性研究

The applicability of laser point cloud registration based on Super4PCS

针对超四点快速鲁棒匹配算法(Super 4-points congruent sets,Super4PCS)对不同特点的激光点云配准的适用性问题,分别对不同重叠率、不同噪声点、不同数据量的点云进行了配准实验,比较分析了算法的配准精度和数据适用性。实验分析得出,Super4PCS算法对于重叠率高于40%、噪声比例低于30%的点云数据,具有较好的配准效果,而在较高精度参数设置下(d<1),点云数据量接近300万时,Super4PCS算法难以完成点云配准。在实际点云配准时可以依据点云数据的特点,判断数据是否适用于Super4PCS算法,通过处理点云数据来满足Super4PCS算法配准的数据条件,得到良好的点云配准结果。

Aiming at the applicability of Super 4-points congruent sets(Super4 PCS)for laser point cloud registration with different characteristics,registration experiments are carried out for point clouds with different overlap rates,different noise points and different data volumes,and the registration accuracy and data applicability of the algorithm are compared and analysed.Experimental analysis shows that the Super4 PCS algorithm has a good registration effect for point cloud data with an overlap rate of more than 40%and a noise ratio of less than 30%.Under the setting of higher precision parameters(d<1),when the amount of point cloud data is close to 3 million,the Super4 PCS algorithm is difficult to complete the point cloud registration.In the actual point cloud registration,it can be judged whether the data is suitable for the Super4 PCS algorithm according to the point cloud data’s characteristics.A good point cloud registration result can be obtained by processing the point cloud data to meet the data conditions of the Super4 PCS algorithm registration.