A Linear-Color Contrast Based Algorithm for Fault Detection of Primary Spring in Train Bogie System
Longkai Liu, Yuanjiang Hu, Shupan Li, Meng Zou, Deqing Huang, Xiaoman Liu · 2020 IEEE 9th Data Driven Control and Learning Systems Conference (DDCLS) · 2020
The primary spring buffers the impact and vibration of track on train. Hence, once the primary spring breaks or shifts, it will cause serious safety accidents. To detect whether the primary spring has shifted or not, the linear-color contrast (LC) algorithm is adopted to locate a marked white line on the primary spring. More specifically, the salient value of each pixel in the image is calculated to binarize the image. Then, by calculating whether the abscissa difference of the white line's contour between the template image and the image to be detected is less than a given threshold, the primary spring in the image to be detected is determined to be shifted or not. By comparing with other threshold processing techniques for image segmentation tasks, including basic global threshold processing, OSTU (is also called maximum between-cluster variance (MBV)) and Sauvola, the LC algorithm achieves the most impressive performance in primary spring fault detection task.