WIT Press

Application Of Image Processing To Maintenance Of Signalling Facilities


Free (open access)

Paper DOI






Page Range

73 - 84




3,505 kb


N. Nagamine & M. Ukai


Signalling facilities are important for ensuring safety and stability of train operation, and, the maintenance of signalling facilities have direct relevance to safe and stable train operation. Remote sensing and database management have increased efficiency of maintenance of signalling facilities. Nevertheless, there still remain many types of inspection work requiring human resources in maintenance, such as the visual inspection of facilities, checking the visibility of signals, in particular. These types of inspections still require a large amount of human resources, are highly expensive, lacking in objectivity, and do not have sufficient accuracy. Therefore, we have developed several inspection methods for the maintenance of signalling facilities using image processing technologies, in order to reduce the cost and increase inspection precision. As one example of the developed methods, we have developed a visibility check method of inspection for obstruction warning signals. The signals are important signalling facilities used at level crossings. In addition, the light of the signal is turned off in normal situations, and blinking in abnormal situations. Therefore, as for the inspection of the signals they must be checked with respect to the direction of light emission and conditions of the weeds around them. However, it is not possible to check the visibility of such facilities during train operating hours, because blinking the signal would require interruption of train service. In this paper, we propose a method that employs near infrared LEDs, checking of the visibility of the signals without the interruption of the train operation, and tracking and detecting specific blinking of the signal correctly. Finally, we report


image processing technologies, signalling facilities, obstructionwarning signal, level crossing, visibility check, near infrared LED