Fig. 1. Comparison of point cloud colours mapped with different data. (a) Mapping point cloud colours with surface variants; (b) mapping point cloud colours with logarithmic surface variants

Fig. 2. Schematic diagram of grid division

Fig. 3. Comparison of rasterised dimensionality reduction images for different data point clouds. (a) Surface variational dimensionality reduction image; (b) logarithmic surface variational dimensionality reduction image

Fig. 4. Point cloud of the rivet area extracted from the ring column enclosing the box

Fig. 5. Specific process and results of point cloud coarse segmentation of rivet region based on image and point cloud mapping. (a) Original image; (b) edge detection; (c) first contour rejection; (d) rivet circle contour fitting; (e) removal of bifurcation inflection points; (f) second contour rejection; (g) rivet circle contour fitting; (h) rivet point cloud extraction

Fig. 6. Rivet unevenness fefinition and calculation model. (a) Definition of rivet unevenness; (b) rivet unevenness calculation model based on three-dimension point cloud

Fig. 7. Flowchart of the rivet hierarchical fitting algorithm

Fig. 8. Rivet layered structure fitting process and results. (a) Point cloud of rivet region; (b) variational Gaussian fitting of logarithmic surface; (c) rivet structure layering; (d) rivet structure fitting

Fig. 9. Standard part and scanned point cloud. (a) Standard part; (b) scanned point cloud

Fig. 10. Algorithm implementation process and result diagrams. (a) Rivet circular contour fitting; (b) rivet region point cloud; (c) rivet region layering; (d) rivet layering fitting

Fig. 11. Comparison results of the proposed algorithm and NV-D algorithm