Defect Detection and Geometric Parameters Measurement of Railway Fastener Based on 3D Linear Laser Sensor

Abstract Rail fasteners secure rail tracks to sleepers to prevent displacement, which makes them critical for railway safety. Regular track inspection is essential as long-term usage leads to fastener defects and track irregularities that require accurate measurement of geometric parameters of fasteners for proper maintenance. This paper proposes a method for fastener defect detection and geometric parameter measurement based on 3D linear laser sensor. Firstly, a 3D imaging system is constructed based on a 3D linear laser sensor to generate RGB-P bimodal data, which includes an RGB depth image and its corresponding point cloud. Secondly, the visual defect is detected and fastener area is located from the RGB depth image of railway track. By mapping the fastener area of RGB depth image to point cloud, the fasteners point clouds are rapidly segmented from railway track point cloud. Lastly, the PointNet++ network segments the fastener point cloud into individual components. Based on the spatial structure of fastener components, the specifications of insulated blocks, thickness of height adjustment pads, and bolt heights are measured. Experimental validation shows 100% precision and recall in visual defect detection. For fastener components with minimum specification differences of 1 mm, the measurement error remains below 0.5 mm. The system achieves a real-time detection and measurement speed of 4.32 km/h, effectively replacing manual inspection for high-speed railway fasteners.