Design Of Large Rail Docking Measurement System Based On Machine Vision

Traditional rail deviation measurement is generally carried out by hands with thetheodolite for angles obtaining, the reading and data processing. The deviation of theprocessed data is entered manually to the hydraulic control system of regulating railposition for the mission of rail docking. Docking rail design with the traditionalmethod that there is labor-intensive, low efficiency and low precision manyshortcomings, since docking rail is larger and far away from the measured target.With the fast development of computer vision and image processing technology, bothdomestic and foreign image measurement technology products have graduallyreplaced the original conventional optical imaging measurement equipment anddeveloped towards multi-functional, intelligent direction. In view of this, a dockingrail deviation measurement method based on machine vision processing is proposedin this thesis. The measurement system hardware mainly consists of camera, cameracontrol cabinet, calibration collimator and measurement targets. The camera would beequipped with high-resolution CCD camera and high-magnification imagingelectronically controlled zoom lens for the completion of the target image capture.Camera control cabinet equipped with high stability master industrial computers andIntel gigabit Ethernet, which is responsible for control of the camera imageacquisition, image processing and communication with master computer. Calibrationcollimator provides the basis for the calibration of the camera orientation.Measurement target provides feature image for deviation measurement system.Measurement system software was programmed by the Visual Studio2010MFCtemplate. Image processing algorithm was programmed by the OpenCV computervision library. The ROI (region of interest) extraction algorithm was designed basedon contrast threshold comparison of the image ranks. A target deviation method basedon the feature circle center coordinates calculated was proposed for characteristics ofthe before and after the target feature image. A camera orientation calibration methodwas studied based on collimator addresses the case where there is a certain deviationbetween the camera actual orientation and the measurement system requirementsbecause of camera composition. Deviation measurement experiment was carried outin a simulated environment docking rail. Analysis of experimental data shows thatmeasurement system precisions meets the design requirements.