改进型点云数据融合的多站组网SVD算法

在大型结构的加工与装配过程中激光雷达的多站组网测试十分常见,但由于点云数据拼接过程需要统一的坐标系,故环境干扰、站位布局导致的转站误差大幅降低了系统的整体测量精度。为了提高多站组网后点云数据融合的面型测量精度,提出了一种改进型奇异值分解算法。该算法在分析站位布局的基础上,通过在多站之间匀差的方式抑制粗大偏差。对目标函数进行了奇异值分解,并通过最优值完成站位的优化布置。实验采用单点精度001mm的MV350型激光雷达,并对6组不同布站情况进行对比分析。结果显示,本算法的点最大误差为00824mm,点平均误差为00214mm,点测量不确定度为00122mm,均优于未规划的测量结果。其测量综合不确定度最接近单机测量综合不确定度,可见,采用本算法对提升转站精度具有一定价值。

Multi station networking SVD algorithm based on improved point cloud data fusion

In the process of processing and assembling large structures,the multi station network test of lidar is very common,but as the point cloud data splicing process requires a unified coordinate system,environmental interference and station layout lead to transfer errors that significantly reduce the overall measurement accuracy of the system.In order to improve the surface measurement accuracy of point cloud data fusion after multi station networking,an improved singular value decomposition algorithm is proposed.Based on the analysis of the station layout,the algorithm suppresses gross deviations by homogenising the differences between multiple stations.The singular value decomposition of the objective function is carried out and the optimal placement of the station is completed through the optimal value.The experiments are carried out with a single point accuracy of 0.01mm for the MV350 lidar,and six groups of different stations are compared and analyzed.The results show that the algorithm has a maximum point error of 0.0824mm,an average point error of 0.0214mm,and a point measurement uncertainty of 0.0122mm,all better than the unplanned measurement results.The comprehensive uncertainty of measurement is closest to the comprehensive uncertainty of single machine measurement.It can be seen that the use of this algorithm is of certain value in improving the accuracy of transfer stations.