Safety assurance of railway tracks is an important task because defects on the railway track may cause trains to be derailed. Unfortunately, there are many fatal train crashes reported in Turkey, Spain and India caused by derailment of the trains. Besides the fatalities, derailment of the trains causes significant economic losses, delay of the timetables and discomfort. Manually investigating the railway tracks are substantially time consuming and expensive. In addition to this, it is human dependent so that the task is prone to error, which may end up without noticing important defects. In this study, a robust and economical method, which automates the investigation of railway tracks, is proposed. The proposed method scans the railway track by a high-resolution optic camera mounted on an UAV. Obtained optic images of the railway track are georeferenced by automatically extracting the Ground Control Points (GCP). The railway track is extracted from the optic image by edge detection. Afterward, any defect on the railway track such as missing bolt or damage on the sleepers and any impropriety of ballasts can be detected by performing image analysis techniques. Advantages of the optic system are its lightweight, higher georeferencing accuracy, and low power consumption. Sony HX300 DSLR camera with 20.4 MP resolution is calibrated on the ground and optical images are obtained at similar imaging geometry. For this reason, flight altitude is fixed as 12 foot above the railway track. The optical camera is mounted on quadcopter, which has 6 kg of maximum take-off weight. The system is tested in Pittsburgh and satisfactory initial test results are obtained. Consequently, site tests showed that the proposed robust and economic system is a good candidate to be implemented for monitoring the railway tracks.
Intracavity image upconversion system with fast and flexible electro-optic image gating based on polarization-frustrated phase-matching for range-gated applications
