Research on Automatic Detection Method of Railway Fastener Defects Based on Image Processing

摘要

With the rapid development of rail transit, the detection requirements for the various components of the track line are getting higher and higher, and relying on manual detection has the disadvantages of high cost and low efficiency. Therefore, it is urgent to study the method of automatic detection of track lines. This paper is based on the development history of computer vision and deep learning detection algorithm in fastener detection. It mainly introduces the related algorithms of positioning and classification, including the "cross" and template matching positioning algorithm; extracting the image direction gradient histogram The graph and the local binary pattern feature are merged, and the algorithm is classified by the support vector machine. At the same time, the convolutional neural network Alexnet architecture is used to extract the generalization characteristics of the fasteners to improve the classification accuracy of the fasteners. Finally, the problems and dilemmas of the existing fastener detection algorithms are discussed.