End Face Positioning Method for Bundled Bars by Using Point Cloud Partition Statistics

Aiming at the difficulty and low efficiency of end face center position measurement in the end face labeling process of bundled bars, a method of end face positioning of bundled bars based on point cloud partitioning statistics is proposed. Firstly, the end face point cloud is pre-processed; then the end face point cloud is segmented using an improved clustering segmentation algorithm, and a method for locating the endface centre of complete single bar and a method for locating the endface centre of defective single bar based on the zoning statistics are proposed to achieve end face center localization; finally, the end face normal vector is used instead of the endface center pose for end face center pose reconstruction. 200 single bar end face point clouds with different pose were collected for the experiment, the time to locate a single bar endface was less than 0.5 s, the position error was less than 3 mm, and the posture error was within 2°. The results show that the proposed method is efficient and can accurately locate the bar endface centre.