Fig. 1. Flow chart of rivet flushness measurement algorithm

Fig. 2. Skin surface condition. (a) Image of skin surface; (b) Canny edge detection

Fig. 3. Two geometric features of contour inflection point

Fig. 4. Contour segmentation based on neighborhood features. (a)(b)(c) Original contours disturbed by noise; (d)(e)(f) neighborhood features at the inflection point of the contours

Fig. 5. Three kinds of neighborhood features at the inflection point of the rivet contour

Fig. 6. Contour segmentation based on neighborhood features. (a)(b)(c) Contour segmentation

Fig. 7. Rivet contour extraction result corresponding to the skin image shown in Fig. 2

Fig. 8. Mapping relationship between camera pixel and three-dimensional point

Fig. 9. Rivet segmentation in point cloud based on image-point cloud mapping. (a) Extraction of rivet contour in image; (b) rivet segmentation in point cloud

Fig. 10. Rivet segmentation in point cloud based on PCL

Fig. 11. Measurement system for rivet flushness. (a) Hardware of measurement system; (b) software interface

Fig. 12. Standard part and its measurement result. (a)Standard part; (b)measurement result of standard part

Fig. 13. Skin part to be measured

Fig. 14. Images and reconstruction results before and after introducing adaptive projection measurement method. (a)(b) Before introducing adaptive projection measurement method; (c)(d) after introducing adaptive projection measurement method

Fig. 15. Rivet contour extraction and point cloud segmentation. (a) Canny edge detection result; (b) smooth contour after segmentation based on inflection point; (c) extracted rivet contour; (d) point cloud segmentation based on rivet image contour