A Robust Rail Positioning Method for Turnout and Non-turnout Sections

Rail positioning is an essential task for railway track defect detection. The turnout is one of the main components of railway tracks. In order to realize the positioning of rails in turnout sections and non-turnout sections, this paper proposes a robust rail positioning method based on double sliding windows. Firstly, extract lines from the preprocessed image. Secondly, lines are discretized into points, and then all points are projected to the x-axis of the image. Finally, a double sliding window method is used to search for the rail area. The experimental results show that the proposed method can effectively identify the rail’s location in the turnout area, and the detection accuracy can reach 99%. What’s more, it has strong adaptability to complex detection environments such as curved rail or broken rail.