At present, the method of two-dimensional image recognition is mainly used to detect the abnormal fastener in the rail-track inspection system. However, the too-tight-or-too-loose fastener condition may cause the clip of the fastener to break or loose due to the high frequency vibration shock, which is difficult to detect from the two-dimensional image. In this practical application background, 3D visual detection technology provides a feasible solution. In this paper, we propose a fundamental multi-source visual data detection method, as well as an accurate and robust fastener location and nut or bolt segmentation algorithm. By combining two-dimensional intensity information and three-dimensional depth information generated by the projection of line structural light, the locating of nut or bolt position and accurate perception of height information can be realized in the dynamic running environment of railway. The experimental results show that the static measurement accuracy in the vertical direction using the structural light vision sensor is 0.1 mm under the laboratory condition, and the dynamic measurement accuracy is 0.5 mm under the dynamic train running environment. We use dynamic template matching algorithm to locate fasteners from 2D intensity map, which achieves 99.4% accuracy, then use the watershed algorithm to segment the nut and bolt from the corresponding depth image of located fastener. Finally, the 3D shape of the nut and bolt is analyzed to determine whether the nut or bolt height meets the local statistical threshold requirements, so as to detect the hidden danger of railway transportation caused by too loose or too tight fasteners.

The non-coplanar lasers on both sides of the rail during full-section rail profile measurement based on line-structured light vision will cause the measured profile to be distorted, resulting in measurement errors. Currently, in the field of rail profile measurement, there are no effective methods for evaluating laser plane attitude, and it is impossible to determine the degree of laser coplanarity quantitatively and accurately. This study proposes an evaluation method based on fitting planes in response to this problem. Real-time fitting of laser planes with three planar targets of different heights provides information about the laser plane attitude on both sides of the rails. On this basis, laser coplanarity evaluation criteria were developed to determine whether the laser planes on both sides of the rails are coplanar. Using the method in this study, the laser plane attitude can be quantified and accurately assessed on both sides, effectively resolving the problem with traditional methods that can only assess the laser plane attitude qualitatively and roughly, thereby providing a solid foundation for calibration and error correction of the measurement system.

The current method of dynamic rail profile measurement involves the installation of a line-structured light sensor at the base of the train. The accuracy of this measurement is influenced by the vertical relationship between the laser plane of the light sensor and the longitudinal direction of the rail (LDR). When a train travels in a straight line, the normal of the laser plane aligns with the LDR. However, when the train curves, the angle at which its wheels connect with the rails causes the laser plane’s normal direction to deviate from the LDR, leading to measurement errors. To address this issue, we propose a method for curved rail profile measurement using a dual-beam laser to correct these errors. This method involves generating an auxiliary 3D rail reflecting the LDR and a virtual 3D rail reflecting the normal direction of the laser plane from the cross-section image of the dual-beam laser. An optimization function is then formulated to determine the optimal auxiliary plane (optimal-AP) by analyzing the alignment or intersection between the auxiliary and virtual 3D rails. Distorted contour points are projected onto the optimal-AP to rectify errors. Experiments validate the accuracy and effectiveness of this proposed method. The results show that, regardless of pitch or yaw movement between the laser plane and the LDR, the error in measuring corrected profile wear remains consistently below 0.10 millimeters, thereby meeting the accuracy standard for rail wear measurement. This approach rectifies measurement errors in curved rail profiles from a 3D perspective, ensuring accurate measurements even under complex working conditions. It also provides a valuable reference for error analysis and improving dynamic rail profile measurement accuracy.

The smooth area on the rail surface causes abnormal exposure in the laser section image, resulting in measurement errors of the rail profile. To address this issue, a novel rail profile measurement technique based on polarization fusion imaging is proposed. A polarized camera is utilized to acquire the four-directional polarization component images, Stokes parameter images, linear polarization angle images, and linear polarization degree images of the rail laser section. A polarization image data fusion algorithm based on Segmented Random Sample Consensus (S-RANSAC) is designed using these images as data sources, and the optimal rail profile fitting curve is obtained. Experimental results demonstrate that the proposed method can still obtain accurate and effective rail profile data in regions where traditional methods fail to capture profile data. Compared with the traditional method, the measurement error of the rail profile is reduced from 0.137 mm to 0.081 mm, and the measurement accuracy is improved by 40.9%. Evidently, this method avoids data loss in key areas of the rail profile caused by local underexposure, thus significantly enhancing the measurement accuracy. This method can provide a valuable reference for high-precision measurement of the rail profile under complex working conditions.