Study On Measurement Method Of Geometric Parameters Of Metro Contact Rail

Contact rail power supply is a common power supply form in metro.Its geometric parameters have great influence on the power supply performance of contact rail.Contact rail contact with collector shoe for a long time will increase the probability of geometric parameter deviation from the design value,and then affect the normal power supply.Therefore,it is necessary to regularly check the geometric parameters,while the traditional measurement methods have problems of large error and low efficiency.According to the requirement of highefficiency maintenance of the contact rail of metro operation company,the method of monocular camera with line structure light is proposed to detect the geometric parameters of contact rail.Aiming at the problems existing in the accuracy,efficiency and adaptability of the detection method,this thesis studies and improves it,designs a set of high-efficiency,high-precision and strong adaptability detection system,and finally tests the measurement function of the system through experiments.The main contents of the thesis are as follows:(1)The calibration method of internal and external parameters of camera is studied and improved,and the high precision imaging mathematical model is established.The imaging principle of the camera is analyzed.The calibration of the camera internal parameters and the optical plane parameters is carried out by the method of concentric ring calibration target and the indirect feature extraction method.The calibration results show that the two methods can effectively improve the calibration accuracy.The detected point set is transformed from the world coordinate system to the orbit coordinate system by ICP point set registration algorithm,and the unity of the detection system and the contact rail geometric parameter reference coordinate system is realized;(2)The method of morphology and edge detection is proposed to extract the main body of the light bar,to solve the problem of noise influence of the structure light in the image of the dark light and the light environment,and improve the applicability of the detection system in different environments.The noise generated by image background light source is removed by the closed operation in morphological method;For the scattering noise generated by the light stripe,the image denoising method based on edge detection operator is adopted.By comparing the denoising effect of several edge detection operators,Sobel edge detection operator with the best noise reduction effect is selected to remove the scattering noise,and finally the light bar with good shape is obtained;(3)The traditional Steger light stripe feature center extraction algorithm is improved to improve the extraction efficiency of light stripe center line.Steger algorithm based on Hessian matrix has high extraction accuracy,but it has the problem of large amount of calculation.Based on the original algorithm,this thesis proposes a method of extracting structured light center by introducing light skeleton and IIR Gaussian filter.This method can locate the center region of light stripe,and simplify the convolution process by using Gaussian filter recursive algorithm,so as to realize the fast extraction of light stripe center.The experimental results show that the extraction efficiency of the improved algorithm is improved by more than 57%;(4)The feature points of different geometric parameters are extracted by slice statistics and constraint line method.The distribution of feature points of geometric parameters on the extraction center line is analyzed.For the measurement positions with more feature points,the coordinate interval with the largest number of slice statistical elements is analyzed.The geometric parameters are represented by the coordinate average of the geometric feature direction of the feature points in the interval;For the measurement location with less feature points,the intersection of the pre-set constraint point line and the fitting curve of the point set within the threshold range is taken as the feature point,and the geometric parameters are represented by the coordinate value of the geometric feature direction of the feature point.In the global ranging test,the results show that the geometric parameters obtained by the above method have high accuracy and stability.