A Point Cloud Connectivity Reduction Algorithm Based on Distance Thresholds

In order to improve the accuracy of target identification and positioning in the unstructured environment of industrial sites, a point cloud connectivity reduction method is proposed for calculating the near-neighbor average Euclidean distance of the workpiece point cloud boundary K to address the connectivity problem between stacked workpiece point cloud data. Firstlyˈthe boundaries of the point cloud data will be extracted; secondly, the average Euclidean distance from the point cloud boundaries to the non-boundary points of K nearest neighbors will be calculated. The characteristics of the average Euclidean distance distribution at different points on the boundaries will be analyzed, divides the boundary points into junction boundary points and non-junction boundary points, and sets a reasonable distance threshold as a screening condition to remove the junction point clouds and retain the original boundary information to the maximum extent will be set. Finally, the identification rates of three pipe joints are analyzed in the constructed experimental platform. The experimental results show that the method proposed in this paper can effectively reduce the connectivity of point clouds and improve the success rate of subsequent identification, with the average success rate of target identification increasing from the original 90.3% to the present 96.3%.