A laser stripe segmentation algorithm for wheel tread profile of rail vehicles under ambient light interference

In the measurement of the wheel tread in rail vehicles, line laser vision measurement technology has a good application prospect. However, the intensity and location of the ambient light will constantly change in the actual application scenarios. Traditional laser stripe segmentation algorithms often fail to produce accurate results, leading to decreased measurement precision in wheel tread. To solve the problem, a segmentation algorithm for laser stripes was proposed. Firstly, the SSR algorithm and frame subtraction were utilized to remove the background noise. Then, the OTSU method was used for the preliminary segmentation. After that, smoothing Images and reducing noise were performed with geometric mean filtering and morphological closing. Finally, the segmentation function which was based on the gray scale distribution characteristics of each region of the image was established to achieve the accurate segmentation of laser stripes. Laser stripe segmentation experiments, laser stripe segmentation comparison experiments, and wheel tread geometry extraction experiments were designed and conducted under the ambient light interference. The experimental results show that the segmentation success rate of the proposed algorithm is not <90.625 %. The proposed algorithm has a superior segmentation effect compared to other algorithms. The proposed algorithm can improve the measurement accuracy. For flange height measurement, the mean error decreased from 0.298 mm to 0.161 mm, and the standard deviation decreased from 0.600 to 0.548. For flange width measurement, the mean error remained constant at 0.200 mm, and the standard deviation decreased from 0.681 to 0.536. Under the condition that the ambient light intensity is in the range of 37lux∼1050 lx and the laser power is not <50mW, the proposed algorithm can better realize the adaptive segmentation of laser stripes.