基于三维激光点云的零件表面缺陷检测

Parts Surface Defect Detection Based on 3D Laser Point Cloud

为了获取零件表面缺陷的精确尺寸,提出了一种基于三维激光点云的零件表面缺陷检测方法。首先使用三维激光测量仪对受损零件进行三维测绘,对测得的原始点云经过离群点去除、RANSAC平面分割等操作后得到零件点云。通过基于SAC-IA的粗配准和基于Huber损失函数优化的ICP精确配准算法,将缺陷零件的点云和完整零件的点云(从CATIA中导出)进行配准,使2个点云对齐。最后利用kd-tree加速的最邻近查找算法得出零件缺陷部位点云。分别使用了含破洞、凹陷的零件验证算法的有效性。实验结果表明:论文提出的方法对零件表面缺陷的3D尺寸提取精准。为后续开展零件修复工作打下良好的基础。

In order to obtain the exact size of parts surface defects,a surface defect detection method based on 3D laser point cloud was proposed.Firstly,a 3D laser measuring instrument was used for 3D mapping of the damaged parts.The point cloud of the damaged parts was extracted from the raw point cloud data using the outlier point removal algorithm and the plane segmentation method based on RANSAC.The SAC-IA algorithm was employed for coarse registration and the ICP algorithm based on the optimal Huber loss function for accurate registration.So two sets of point clouds data for the defective part and the complete part were a⁃ligned.Finally,the point cloud of the defective regions was obtained by using the nearest neighbor search algorithm accelerated by kd⁃tree.Parts with holes and depressions were used to verify the effectiveness of the algorithm.Experimental results show that the method proposed in this paper can accurately extract 3D dimensions of surface defects of parts,which lays a good foundation for the following repairment of the parts.