弱偏差特征过滤的建筑物焊接连接构造点激光雷达特征提取  

Lidar feature extraction of building welded joints based on weak deviation feature filtering

作　　者：江浩 齐慧芳 JIANG Hao;QI Huifang(Guangzhou No.1 Decoration Co.,Ltd.,Guangzhou 510060,China;Guangzhou Vocational College of Technology&Business,Guangzhou 511442,China)

机构地区：[1]广州市第一装修有限公司,广东广州510060 [2]广州科技贸易技术学院,广东广州511442

出　　处：《粘接》2023年第8期134-137,共4页Adhesion

基　　金：广州市高校科研项目(项目编号:202235418)。

摘　　要：在扫描建筑物焊接连接构造的过程中,由于构造点区域缺少明显的特征,易形成弱偏差干扰,提出一种在弱偏差特征过滤下,应用激光雷达扫描的特征提取方法。利用激光雷达扫描中的阵列分析法,采集建筑物焊接连接构造的点云数据,利用KD-Tree模型剔除点云集合中的弱偏差特征噪声数据;采用邻域方向分布法,在被跟踪的构造点集合中标记出倒数的2个点,并搜索初始集合中相连的有向线段,将最小扫描角度的线段端点加到特征集合中。结果表明:该方法提取的位置精度高,区域匹配度最高值可达到95%。During the process of scanning the welding connection structure of buildings,weak deviation interference is easily formed due to the lack of obvious features in the construction point area.Therefore,a feature extraction method using LiDAR scanning under weak bias feature filtering is proposed.The array analysis method was used in LiDAR scanning to collect point cloud data of building welding connections,and the KD Tree model was used eliminate weak deviation feature noise data from the point cloud set;the neighborhood direction distribution was used to mark the two reciprocal points in the tracked construction point set,and search for the connected directed line segments in the initial set,adding the endpoint of the line segment with the minimum scanning angle to the feature set.The experiment shows that the method has high accuracy in extracting positions,and the highest value of region matching in the extracted results can reach 95%.

关 键 词：激光雷达扫描 弱偏差特征 焊接连接构造 压缩感知 KD-Tree模型 向量点乘法 轮廓点跟踪 

分 类 号：TN249[电子电信—物理电子学] TQ573.8[化学工程—精细化工]