Research On 3D Detection And Application Of High-speed Railway Turnout Based On Structured Light Technology

With the rapid development of high-speed rail, the detection requirements of the components of the railway, including the detection accuracy, detection speed and detection method, are continuously improved. Many advanced detection technology and equipment have been used in the detection of railway in order to adapt to the new testing requirements. Because of its simple structure, low system operating requirements and adaptation of a variety of inspection tasks, machine vision become one of the hottest areas of rail detection research in recent years. High-speed turnout is a vary important part of the high-speed railway and its surface structure is complicated and higher precision. So it has important practical significance to research on and explore the application of machine vision in high-speed turnout detection.In this thesis, the key technologies and application in high-speed 3-D turnout detection of binocular structured light which is based on machine vision is deeply researched. A detection system of high-speed turnout based on binocular structured light is preliminary designed and achieved. Binocular structured light which combines the advantages of binocular stereo vision and the grating structure technology is commonly used in 3-D measurement. And then, for the characteristics of Turnout, the hardware configuration and software modules of detection system of high-speed turnout is designed.Research on the calibration method of binocular structured light system, and achieve a quick calibration and experimental verification based on planar target. Considering the principle of DLP projector and measurement model of the system, a method for optimizing the viewpoint is developed to overcome the uncertain effect result from manual operation in binocular structure light detection process. Based on the optimization theory, a detection path planning program of the system is proposed, and then a measuring trolley in order to achieve the detection path planning is designed. In the data collection process, for measuring characteristics of the system, the method of multi-view mosaic is improved reasonably. At last, deal with the point cloud data of high-speed turnout which was collected by the detection system using Geomagic Qualify. After pretreatment and analysis of point cloud data, the detection report of high-speed turnout is generated.