Wheelset Tread Detection Design Based On Machine Vision

As the rapid development of railways and urban rail transit in the current era,a large number of trains are running on the track around the world,and the hidden safety hazards are increasing gradually.As a part of the train which directly contacts with the track,the wheel set tread will produce a certain amount of wear every day.Over time,this kind of wear has a certain impact on the running safety of the train.The mechanical size of the wheel set tread is measured as an important step according to the rules to ensure the train's safe running.This paper designs a wheel set tread automatic detection system based on advanced machine vision and image processing technology for the low precision and low efficiency of traditional manual wheel alignment detection methods for rail transit trains.The system is arranged beside the track,and a laser light source is used to emit a line laser to the wheel,then the image of the light band is collected by the camera.Then the detected parameters are calculated using image processing techniques based on camera calibration results.In terms of algorithm and software,the traditional denoising method is abandoned to use the discrete cosine transform denoising method to denoise the image.The method of improving the edge detection parameters of the traditional edge detection makes the operator can automatically calculate the optimal parameters and enhance the stability of the processing.In the thinning process,the traditional method is optimized for the characteristics of the wheel tread detection,so that the thinning error is reduced to a single pixel or less.Then the camera is calibrated and the corresponding result is obtained.Finally,the image processing results are combined with the camera calibration results to calculate the detection parameters,and the algorithm is written into a program with a graphical interface,and connected to the database to save the results.Experiments show that the system and software can quickly detect a large number of wheel wear automatically,including the thickness and height parameters of the rim.And the tread curve form can be displayed in the interface.It can be introduced into database for storage at the same time,reduced the loss of human resources and improved the efficiency.At the same time,compared with the traditional algorithm,the improved algorithm improves the detection accuracy,and has a certain application significance in the field of train wheel tread detection.