Measurement System Of Dynamic Envelope Curve Of High-speed Vehicle Based On Monocular Vision

In recent years,with the rapid development of railway construction in C hina,the train speed has also been greatly improved,which facilitate the travel of people while put forward higher requirements on the traffic safety.Dynamic envelope curve of high-speed train is defined as the maximum limit outline affected by a variety of adverse factors while the train is running.To precisely obtain the dynamic envelope curve of high-speed train is an important guarantee for safe driving.For this,a dynamic gauge measurement system based on binocular vision,which get the maximum limit outline of the train by catching the information of the measured body using visual sensors,is designed by School of Precision Instrument and Opto-Electronics Engineering,Tianjin University.The system could meet the needs for measuring dynamic envelop curve of high-speed train.However,some shortcomings exposed afer several field experiments,such as large amount of data collection,the complexity of the calibration process,poor adaptability,poor portability and so on.Therefore,to guarantee the traffic safety of high-speed train,it is necessary to develop a measurement system with higher efficiency,better adaptability,more meeting the measurement requirements.Compared with binocular stereo vision or multi-view stereo vision,monocular vision measurement technology has the advantage of unnecessary image feature match,less amount of data collection,low cost and so on.Therefore,it has an extremely vital significance to apply the monocular vision measurement technology in the high-speed train running attitude measurement,to complete the detection of high-speed train dynamic envelope curve.According to the actual measurement requirements and the problems we found in field measurements,in this paper,we apply the monocular vision measurement technology to the field of high-speed train dynamic envelop curve measurement and design the dynamic envelop curve measurement system based on monocular vision.The main jobs are described as follows:1.Starting from the actual measurement demands,we find out the advantages and disadvantages of the previous dynamic envelop curve measurement system,and then propose a new method of high-speed train dynamic envelope curve measurement system based on monocular vision.2.We build the model of monocular vision measurement system and set up system hardware platform according to measurement requirements.In order to improve the adaptability,a zoom lens is built-in the visual sensor.3.In order to calibrate parameters of system rapidly in the field,a high-accuracy 3D target with fast reconstruction is also designed.According to the requirements of visual sensor intrinsic parameters calibration and external parameters calibration,the total linear constrained optimization algorithm is adopted;The way using virtual feature points solved by beam feature points to calibrate the high power line laser is also proposed.4.Analyzes some common algorithmics to extract stripe center points.According to the actual measurement requirements,we conduct experiments to test their noise immunity.Finally,we choose the Steger algorithmic to extract stripe center points based on experimental results.5.Calibrate the system parameters fieldly,and conduct the measurement experiment,which proved the feasibility of the system and evaluated the system accuracy.Also,simulate complex scenario,the measurement experiments are proved the adaptability of the system.Experimental results show that when the system are in the range of 8~16 meters away from the measured body,system accuracy can reach up to ±2mm,which meets the requirements to measure the dynamic envelope curve of high-speed vehicle.