Rail defect detection with real time image processing technique

Abstract

Rail defect detection has a great importance for railway transportation. Because faults on the rails cause to problems such as the cost, the disruption of transportation and the safety. In this study, based real-time video processing algorithm used Morphological feature extraction is recommended to detection defects on railroad. The rail is detected by applying Hough Transform and image processing techniques to rail images obtained from the real time camera. Features of the detected rail images are extracted with applying morphological operations and defected regions are identified. Headcheck, fracture, scour and undulation failures are detected through proposed method. In this study, results with low uptime and high success rate are acquired by performing all steps of proposed algorithm to the images of rail taken under different lighting and direction. © 2016 IEEE.