Broken Railway Fastener Detection Based on Adaboost Algorithm

The detection of broken railway fastener is important to ensure the safety of the railway transport. This paper proposes an efficient method to detect and recognize the broken fastener with complex ballast railway images. Firstly, a from-coarse-to-fine strategy according to the sleeper region´s gray and gradient characteristics is used to position the fastener, then the Haar-like feature set according to the fastener´s geometrical characteristics is introduced. Finally, the fastener state is recognized by the AdaBoost-based algorithm. The method can detect fastener effectively and automatically with high positioning and recognizing accuracy and need not manual intervention. The experiment showed that the detection rate is satisfactory.