摘要
针对飞机蒙皮表面扫描点云中铆钉自动提取的问题,对点云密度计算、基于区域增长的点云聚类、主成分分析投影、随机一致性采样圆拟合等方面进行了研究。对要测量铆接质量的机体结构区域进行了高精度三维激光扫描,获取了相关区域三维点云信息;遍历了所有点云,计算了每个点的局部自适应密度,筛选出了高密度点云区域;进一步对传统的点云密度计算进行了改进,提出了一种局部自适应密度计算方法,利用区域增长的聚类方法,提取出了单独的铆钉轮廓点云,增加了铆钉轮廓点云的显著性;最后针对单个铆钉轮廓点云,采用主成分分析投影技术处理,并进行了随机一致性采样圆拟合。研究结果表明:该算法能有效提高铆钉轮廓的显著性,实现铆钉轮廓点云的自动提取。
Aiming at the automatic extraction of rivets from point clouds scanned on aircraft skin surface,the density calculation of point clouds,clustering of point clouds based on region growth,principal component analysis projection and random consistent sampling circle fitting were studied.High-precision three-dimensional laser scanning was carried out on the body structure area to measure riveting quality,and the three-dimensional point cloud information of the relevant area was obtained.Then all point clouds were traversed,and the local adaptive density of each point was calculated to screen out the high-density point cloud area.The significance of rivet contour point cloud was further increased,the traditional calculation method of point cloud density was improved,and a local adaptive density calculation method was proposed.A single rivet contour point cloud was extracted by using the region growing clustering method.Finally,for a single rivet contour point cloud,the principal component analysis projection was used.After technical processing,random consistent sampling circle fitting was performed.The results indicate that the algorithm can effectively improve the saliency of rivet contour and realize the automatic extraction of rivet contour point cloud.
作者
李红卫
LI Hong-wei(AVIC.XI’AN Aircraft Industry(Group)Company LTD.,Xi'an 710089,China)
出处
《机电工程》
CAS
北大核心
2020年第6期719-723,共5页
Journal of Mechanical & Electrical Engineering
基金
南京航空航天大学航空创新基金资助项目(1005-YQR17001)。
关键词
三维点云
铆钉
点云密度
随机一致性采样
3D point cloud
rivet
point cloud density
random consistent sampling
