Study On The OCS Inspection System Based On Image Processing

With its advantages of stronger transport capacity, lower operating costs, less energy consumption, smaller environmental pollution and so on, the electrified railway is universally concerned and has became the railway development orientation. The overhead catenary system, as one of the most important components in the electrified railway, its state pros has been taking great effects on the safe operating of the electrified railway. To ensure that the electrified railway would operate safely, it is essential to inspect the catenary system periodically and keep it in good condition. At the same time, in order to accumulate experiences of design, construction and maintenance, to increase the catenary system overhaul efficiency and save operating costs, improving the detecting techniques level has become an inevitable requirement and guarantee for the rapid development of the electrified railway.At the end of 20th century, the Contact-Style Detection System has become more and more mature and widely used both at home and abroad. However, due to technical complexity, high cost and some potential safety problems, this technology can not improve better. The Non-Contact-style Detection System, which installed the detection equipment on the roof and completely disengaged with high-voltage equipments, making technology and equipment relatively simple, security features improving and costs dropping.In this dissertation, the problems under High-speed detection are mainly researched. The Non-Contact-style Detection System divides the camera-collected images into multi-thresholds. Firstly, using the adaptive threshold and twice binary approaches to pre-process the images. Then, using tracking distinguishes techniques to eliminate interference targets. Finally, get needed geometric parameters. This makes the target recognition quality and parameters accessing efficiency obviously increased. Building on this, the system optimizes the setting of car-roof detection unit. It puts forward new ideas of the thresholds dividing, multi-objects identifying, lighting sources control and breaks relaying etc.