Research on surface defect detection system and method of train bearing cylindrical roller based on surface scanning

Aiming at the detection of surface defects of train bearing rollers, a set of surface defect detection systems for bearing cylindrical rollers based on machine vision is designed in this paper, and the vibration response of the roller during the rolling process is simulated and analyzed. The surface image of the cylindrical roller is obtained by using the area scan camera, and the region of interest is extracted by using edge extraction and template matching algorithm to obtain the cylindrical roller image without background interference. The roller cylinder images are spliced by feature matching and regional fusion methods, and the roller surface defects are extracted by OTSU threshold segmentation, Canny operator edge detection and morphological processing, so as to mark the defects for distinction and identification. The method achieves the purpose of quickly identifying the roller surface defects and rejecting the cylindrical rollers with defective surfaces in train bearings. It provides a theoretical basis for the automation and intelligence of the cylindrical rollers of train bearings in the process of maintenance and repair.