Research on 3D Point Cloud Reconstruction and Defect Recognition Algorithm for Aircraft Body Rivets

Traditional defect detection methods for aircraft body rivets heavily rely on image recognition, which is often hampered by suboptimal image quality. This paper proposes a novel approach that leverages 3D point cloud data to comprehensively reconstruct the three-dimensional shape of aircraft body rivets, offering an intuitive representation of their shape while enabling defect identification. This paper focuses on the research of the 3D reconstruction and defect recognition methods for rivets and introduces a reconstruction algorithm that first generates localized regions and then fills in the remaining areas. Initially, density clustering is used to extract the rivet regions. When reconstructing the rivet model, priority is given to generating parts with high regularity. Subsequently, the remaining patches are generated based on the minimum cycles extracted using a constrained depth first search algorithm. Furthermore, the reconstructed triangular patches' angle distribution intervals are provided. The results of case studies demonstrate the effectiveness of this method, yielding superior reconstruction outcomes and markedly enhancing the quality of triangular patch angles. This approach exhibits promising applications in 3D point cloud reconstruction and defect recognition within the domain of aircraft body rivets.