The Design Of Railway Components Image Detection System

The state detection for railway components has become an important method to keep security coefficient of railway transport. The papaer uses computer,image processing and pattern recognition technologies to replace the traditional human inspection,which could be more effective as well as more economical. In the meantime, it could improve the efficiency and accuracy for the inspection.According to the requirements of circumstance and performance,selecting the suitable key equipments to build the railway inspection hardware platform.Then design the software implementation program for the defect identification system on that hardware foundation, and foucus on the analysis and improvement of the rail surface defects detection algorithm and fasteners missing detection algorithm.In this part, it mainly includes two parts:the target area extraction of railway image and the defect feature recognition.The target area extraction of railway image is a necessary prerequisite for defect detection.The paper raises a method of adaptive binary image projection based on peak district detecting to thr rail surface region.For the fastener deletion region extraction,using isotropic Sobel vertical edge detection operator and Hough transform to track binary image and extract the edges which meet the threshold requirement, and then search the position of fastener region by using the vertical edge line as reference.On the basis of the target area extraction, defect features recognition is researched. For the rail surface defect area, internal point hollowed algorithm and chain code tracking algorithm are adopted to extract and store the defective target contour information, calculate the characteristic parameters of defects in the target area, perimeter and aspect ratio. According to these parameters the rail surface transverse cracks, longitudinal cracks and peeling off the preliminary classification is achieved.For the recognition of fasteners missing,an improved estimation algorithm based on the direction of the gradient field is proposed. By using this algorithm, the platform could offer a clear picture of fattener texture towards. In order to achieve the fastener status classification, the paper adopts pattern template matching algorithm and a algorithm supporting vector classification.Using Matlab to verify the reliability and validity of the detection algorithm.and the results show that the above algorithm can quickly and accurately detect the rail surface defects and fasteners missing, which can satisfy the technical requirements, such as no image missing, online coarse identification and offline precise identification.Finally,combing Open CV image processing module,with Visual C++6.0writing system software to achieve the automatic detection of the state detection for railway component.