| Since the first appearance of railways,China and the world have paid more and more attention to the development and maintenance of railways.Wide and narrow joints,as the weakest equipment in the track,the traditional detection methods can no longer meet the needs of accurate and efficient detection.Therefore,this paper takes rail board images as the research object,combined with image processing and seam detection,and proposes a machine vision-based rail board seam detection system,which can accurately detect pictures containing wide and narrow seams.The main work of the thesis includes the following aspects.First of all,in the image preprocessing,because of the uneven illumination of the track board image,including various types of noise and other shortcomings,through theoretical comparison analysis and experimental simulation,several algorithms commonly used for image enhancement,filtering and binarization are compared.Then in the straight line detection link,the improved Hough algorithm is used to perform straight line detection on the processed image,and the pixel characteristics of the expansion joint are cleverly combined to repair and effectively extract the detected straight line.Compare the extracted straight line with the reference line of the previous picture to determine whether the detection is accurate.By updating the successful baseline and combining the periodic features of the expansion joints,the pictures containing wide and narrow joints are selected.Finally,Microsoft Visual Studio 2017 and Open CV are used to develop the detection system of the track plate,and the accuracy and stability of the software are tested.The experiment proves that the algorithm in this paper has good stability and high accuracy.Based on the experimental data set,the accuracy is as high as 99%,which has good practical application value. |
