An Online Inspection System for Critical Train Roof Components Based on the Plane Array Camera

For the online monitoring demand of urban train roof condition, an online inspection system for critical train roof components based on the plane array camera is proposed. First, the roof images are continuously collected by the plane array camera. Second, separate images are stitched to obtain the complete roof image by the algorithms of image pre-processing, feature detection, feature matching, and image fusion. Then, the critical roof component is identified and located based on the improved YOLOv5s model. Final, the system performance is analyzed by using the on-site real-time image. The experimental results show that the system has good roof image stitching effect and can accurately identify critical components, which can be applied to practical engineering.