Research On Machine Vision Detection Technology Of Rail Surface Defect

Railway transportation has a large proportion in the transportation industry because of its characteristics of large volume,sustainability and environmental friendliness,and railway transportation is very important to the economic development of our country.The rapid development of railway transportation with high speed and high load is easy to produce defects on the rail surface.The surface defects of the rail not only affect the comfort of the train,but also cause great hidden the train when it is serious.Under the background of increasing train speed,both in precision and speed,the requirement of rail defect detection is higher and higher,and the traditional rail defect detection technology can not fully meet the high requirements of today’s railway defect detection.In the environment of continuous development of detection technology,more and more technologies are used in rail defect detection,and machine vision is widely used in detection technology because of its advantages of non-destructive,automatic and low cost.This paper uses machine vision technology to carry on the related research to the rail defect detection,the research content is as follows:(1)This paper introduces the background significance of rail defect detection research,summarizes the common types of rail defects,summarizes the characteristics of rail defect detection technology and related equipment,especially the advantages of machine vision related technology in detection.On the basis of this,this paper carries on the following related content research and the structure arrangement.(2)According to the imaging characteristics of rail tracks in complex environment,the hardware structure of rail defect detection equipment is designed.The equipment considers the incident angle of light source,the intensity of light source,the distance and so on,and improves the quality of the collected image.the designed rail defect detection equipment is tested experimentally on the experimental platform,and the ideal rail image can be obtained.(3)Based on guided filtering and maximum entropy threshold method,a new algorithm for detecting railway defects in complex background is proposed.According to the difference of variance between the defect region and the non-defective region,this method analyzes the characteristics of guided filtering filter.The improved guided filter is used to smooth the defect region and the original image.The experimental results show that the segmentation effect is better than the traditional image segmentation algorithm,and the improved maximum entropy threshold segmentation method is independent of the background entropy,which makes the calculation time better than the original maximum entropy.(4)According to the change of gray value on different scales of railway track image,the y direction filtering algorithm of railway track surface defect image is proposed.The pixel gray value is introduced in the method of detection of rail surface defects filtered along the y direction.The filtering method only smoothes the y direction and determines the smoothness degree according to the gray value.The experimental results show that the rail defects can be effectively detected by image difference,the time complexity is low and the execution time is 40.6 ms.(5)A defect edge extraction algorithm for rail surface is proposed.firstly,the rail image is de-noised,then the defect edge is smoothed according to the gradient.the size of the gradient determines the filtering degree of the edge.the larger the gradient,the greater the smoothing degree.then the smoothed image is distinguished from the denoised image.finally,the raida criterion is used to segment the image to extract the defect edge.Experimental results show that the proposed algorithm for edge extraction of rail surface defects effectively reduces the interference in complex environment.