Research On Automatic Detection Of CPⅢ Of High-speed Rail Based On Two Line Scan CCD Cameras

With the rapid development of economy and society, the transportation networkdeveloped increasingly. High-speed rail as an important part of the transportation network hasalso made extraordinarily rapid development, which makes how to safeguard the security ofhigh-speed rail has become a very important issue. At the present stage, the measurementtechnology of high-speed rail is not developed enough, traditional static detection methodscan not meet the standard requirements of the modern fast railway maintenance and repair.Therefore, there is an urgent need for a high-precision and highly efficient motion detectionsystem to complete the detection of high-speed rail track, in order to safeguard the security ofhigh-speed rail.Nowadays, with the computer technology continues to evolve, photogrammetrytechnology has been widely used in engineering. As a different from traditional methods, thebinocular stereo photogrammetry is a non-contact measurement methods, it has accurate,rapid and real-time characteristics. Therefore, the binocular stereo photography used in thefield of detection of high-speed rail track has very broad prospects for dynamic testing ofhigh-speed rail track. The main research content of this paper can be summarized as thefollowing aspects:1. CP Ⅲ characteristics for high-speed rail, using binocular stereo photogrammetry anddigital image processing technology, combined with accuracy, high efficiency, low costof the non-contact measurement technology feature, design the CP Ⅲautomatic detectionsystems of high-speed rail based on line-scan cameras. Fully discussed and detaileddescription the structure from both hardware and software parts of the detection system.2. On the basis of the binocular stereo photogrammetry principles and line-scan camerasimaging model, to build the photogrammetric system based on two line-scan CCDcameras. Research and analysis of the calibration of the line-scan CCD camera andexperiments to verify its accuracy.3. Using gray stretch and median filtering two kinds of digital image processing algorithmsto carry out image enhancement of OpenCV for the acquired digital image obtained, greatly reduces the noise in the image and projecting the CP Ⅲoutline feature.4. Through the template matching and segmentation to obtain the target area of the CPⅢcontrol points, using the least squares method to calculate the coordinates of the CPⅢcontrol point center. At last, calculate the coordinates of the CP Ⅲcontrol point in cameracoordinate system according to three-dimensional space and transform relations of thetwo-dimensional image from calibration. And with the mileage counting function, toachieve CP Ⅲautomatic detection.5. Through the large number of the CP ⅢDynamic detection experiments, analyzeexperimental data to verify its ability to suffice the accuracy requirements.6. Finally, summarizes the results obtained of this paper. And put forward the insufficientand further improvement of this system programs.