| Turnout detection is an important branch of railway detection,which is significant to the safety operation of trains.Routine detection is attained by manual detection.Manual detection is inefficient and labour-intensive and low-accuracy.With rapid development of railway in China,manual detection is not meet high-speed and heavy-duty and long-range railway detection.The research of turnout detection devices started later in China.To improve the quality and efficiency of turnout detection,it is very important to explore kind of detection device which has characteristics of high-speed and high-accuracy and non-contact.The main works of this paper is research and design on the application of image processing algorithm in switch rail detection.Image processing algorithm is applied to realize the multi-section detection of switch rail by non-contact and static method.In switch rail detection,odometer is designed and manufactured to accurately location the detection section.Based on Zhang calibration method,a new method is proposed to improve the accuracy of laser plane's spatial position calibration.According to the condition of switch rail detection,composition of multichannel extraction is proposed to full and accurately extract laser stripes.A method is demonstrated to extract accurate sub-pixel center,which is completed by rough extraction and accurate extraction.Based on improving sub-pixel center extraction algorithms,the accurate sub-pixel center can be fast location and extracted in high precision.On the platform of VS2013,the switch rail detection software is designed based on MFC and OPENCV.The software design adopts modularization easy to image capture and image processing and visualization.The testing experimental of odometer and software has been done in laboratory environment and outdoor environment.The test result shows that both the odometer and detection software meet the requirement of railway detection.And the both have high engineering practical value. |
PDF Full Text Request