【摘要】 机器人自动加工复杂曲面之前，需要对待作业目标曲面进行区域划分，以适配不同的工艺参数。针对现有方法分割复杂曲面点云产生的过分割和欠分割问题，提出一种自适应区域生长的复杂曲面点云分割方法。首先，设计基于动态平滑阈值的自适应生长准则，用于复杂曲面对象的点云分割，以减少过分割现象；其次，提出欠分割判断准则，以判定分割后的点云集，过滤出欠分割区域；然后，将欠分割区域重分割，直到不存在欠分割区域；最后，以扫描陶瓷洁具素坯得到的复杂曲面点云为实验对象，测试了提出方法对复杂曲面点云的分割效果。实验结果表明，提出方法的分割精度达92.63%、召回率达98.89%、综合评价指标F1分数达95.66,能有效且完备地分割出复杂曲面点云的各个区域。更多还原

【Abstract】 Before the robot machining complex surface automatically, it needs to divide the area of the target surface to fit different process parameters.Aiming at the over-segmentation and under-segmentation problems caused by using the existing method to segment complex surface point cloud, an adaptive region growing method for complex surface point cloud segmentation is proposed.Firstly, an adaptive growth criterion based on dynamic smoothness threshold is designed for point cloud segmentation of complex surface objects to reduce over-segmentation.Secondly, the under-segmentation criterion is proposed to judge the result point clouds of segmentation and filter out the under-segmented regions.Finally, the under-segmented region is re-segmented until there is no more under-segmented region.Taking the complex surface point cloud obtained by scanning ceramic sanitary ware blank as the experimental object, the segmentation effect of the proposed method on complex surface point cloud was tested.Experimental results show that the segmentation precision of the proposed method is up to 92.63%,the recall rate is up to 98.89%,and the F1-score of the comprehensive evaluation index is up to 95.66,and each region of complex surface point cloud can be segmented effectively and completely.更多还原