Research On Automatic Railway Defect Dection System Based On Image Processing

The railway system,as the main artery of the national critical infrastructure and mass transport,plays an important role in China’s economic and social development. Ensure railway track in goodcondition is the foundation to ensure safety of the railway. Rail is one of the major ra ilwaytechnology and equipment, and the foundation of transportation-For a long time, the monitoringof rail track is inefficient, depending on manual artificial defect inspection, and affecting by thepersonnel experience, responsibility, weather conditions and other factors. With the growth ofhigh-speed railway traffic demand, it’s necessary to develop automatic railway defects monitoringsystems to replace manual inspection.In practical engineering applications, two main parts are needed to be monitored, rails and railfasteners.The corresponding defects can be divided into three categories: uneven light band on rail,missing fasteners and fasteners skew. Based on these practical needs, a set of automatic defectdetection system is designed, which is divided into three parts:(1)The hardware part. With CCD camera installed in the bottom of the high-speed train, theimage of entire rail track is captured while trains traveling, and save using a real-time algorithm.(2)Automatic detection section. After obtaining an image, we have designed a set of algorithmsto achieve the detection of various defects and use the C++language to implement the procedures.(3)Defects playback section. In this part, by manual inspection, the false defects are deleted,leaving the defects which are needed to fix.This thesis made a detailed presentation on the second part of the algorithm, using edgedetection, image matching, symmetry detection methods to achieve the detection and localization ofdefects and use the GPU to accelerate image processing. Test confirmed that the system is able totimely and effectively discover a variety of defects to ensure the safety of high-speed trains.