Fig. 1. Rail fastening detection system. (a) Line-surface measurement model; (b) system structure diagram

Fig. 2. Schematics of algorithm . (a) Center point extraction; (b) light bar grayscale distribution at junction

Fig. 3. Results of centerline extraction using different algorithms. (a) Extreme value method; (b) grayscale center of gravity method; (c) Steger method; (d) proposed method

Fig. 4. Centerline extraction results under different influence conditions. (a)(e) No lighting; (b)(f) LED lighting; (c)(g) normal ambient lighting; (d)(h) presence of stain

Fig. 5. 3D reconstruction results of fastener. (a) Extreme value method; (b) Steger method; (c) grayscale center of gravity method; (d) proposed method

Fig. 6. Establishment of fastener coordinate system. (a) Physical drawing of fastener; (b) circular fitting of upper surface of bolt; (c) fastener detection coordinate system

Fig. 7. Combined classifier model of fastener defect detection

Fig. 8. Inspection for skewness of fastener. (a) Height distribution of point clouds in areas C and D; (b) detection diagram of skewness

Fig. 9. Looseness detection of fastener. (a) Detection principle; (b) surface normal vector estimation; (c) detection diagram of nut looseness

Fig. 10. Data collection devices. (a) Indoor collection equipment; (b) standard block point cloud; (c) inspection vehicle collection equipment

Fig. 11. Fault diagnosis. (a) Nut missing; (b) fastener skewness; (c) elastic strip missing; (d) elastic strip reverse

Fig. 12. Experimental results. (a) Defect detection of fastener; (b) tightness detection of fastener