Methods to Test Integrity of Train Components

To address concerns regarding component integrity and safety detection of trains, and the efficiency of this process, an automated system for train component integrity is proposed. First, the collected images are preprocessed to eliminate image noise. The component position is automatically matched to a pre-established part template. Then the component image is developed using the algorithm of multi-point automatic seed growth. Then, the abnormal region is judged based on the growth condition, where the difference region of the contour feature is analyzed to determine whether the component is intact. The experiment tested the integrity of thirty locations requiring detection in ten images. The detection accuracy for some types of components, such as the iron chain and the tail pipe, exceeded 95%.