Aiming at the limitations and the low automations of the current rivet hole feature extraction method

an automatic identification and features parameter extraction method for rivet hole based on scattered point cloud is proposed. The point cloud boundary points are extracted by the k neighbor point distribution of each point

 and the point cloud blocks belonging to different boundary features are obtained by using Euclidean distance clustering

 and then according to the ellipse fitting result

 the rivet hole boundaries are extracted. The k nearest neighbor search algorithm is used to find the point cloud of the rivet hole boundary neighborhood

 and then the actual positioning plane of the rivet hole is constructed

and the boundary point is projected to the positioning plane to fit the end circle. The experiments show that the method can accurately identify and extract the rivet holes and that the center position extraction accuracy can reach 0.029mm.