Research on aircraft skin rivet detection technology based on the normal vector-density clustering algorithm

Riveting quality is crucial to an aircraft’s overall aerodynamic performance and fatigue life. In order to effectively extract the point cloud of rivet heads and analyze the quality of riveting, this paper proposes a rivet flushness detection method based on the normal vector-density clustering algorithm. First, initial point cloud data sampling is based on normal vectors. Then, the density clustering algorithm is employed to cluster and extract the point cloud of rivet heads. Subsequently, the obtained point cloud of rivet heads is subjected to the random sample consensus algorithm for fitting the contour and obtaining the model parameters of the rivet head. The paper introduces a quality detection metric to describe the flushness of the rivet head. Finally, the proposed method is applied to analyze the skin and theoretical model point cloud data of rivets. The results demonstrate that the proposed method yields small errors and high accuracy compared to theoretical values. The method is further employed for quality detection and analysis of rivet flushness in practical aircraft engineering. A visualization system for rivet flushness quality detection is developed to represent the results visually. This system enhances the intuitive identification of rivet detection outcomes. Therefore, the proposed method holds significant engineering application value in rivet flushness detection.