Study On A Technology For Detecting Brakes Of Railway Wangons Based On Machine Vision

Stopping wangon with brakes is a most common way in domestic railway, and will be the important way for a long period. Nowadays, the abrasion detection of brake is mostly completed by technicians. So hard the work that technicians are easily wore out;which results that there is not only the low efficiency, but also the low precision. Therefore, with the advanced computer technology, communication technology and so on, an on-line and automation equipment for detecting the wangon brakes was designed based on machine vision;which would be more practical and wider prospect in the future.Machine vision is a non-touch detection technology;which makes use of all kinds of advanced appliances to establish platform for achieving even breaking through the vision ability of human being. As the increasing development of the chips manufacture and the improving of manufacture technics and the image processing technology, the machine vision technology will be developed quickly and applied widerly.Based on the machine vision principle, the project was aimed to be an on-line and automation technology for detecting brakes of railway wangons. As the result of trials, the clear and intact images were grabbed by image grabber, which was trigged in time by signal trigger module. Images were processed exactly and efficiently, and the brim curves of brake were obtained accurately, which are used to calculate the thickness of brakes and the position of the least thickness.There is a preceding trigger signal module used to trigger the image grabber to grab the coming brake image in time, which is composed of several high speed railway sensors according to the working scene and condition of wangons. The image grabbing module is made up of lights, lens, CCD camera, image grabber and industry computer, etc;which is used to grab all the brake images of one wangon. And the thickness value is calculated by image grabbed with the system calibratation.According to the features of images, there are the image processing arithmetics appled as follow. At first, the image is restored by wiener filter to reduce the blurring arose by the brake-compare-to-camera motion. Then, the canny edge operator isapplied to extract the edge of objects in image. According to the distinct feacture the brake comparing to other objects, such as length and slop of arc, the relative position between brake and wheel, and the distant between two edges of brake, the two edges of brake is extracted on the basis of the cluster analysis principle finally.Before measuring the thickness of brake, the image is converted from field format to frame format, it means that the size is doubled in vertical direction. Then, the edge curves of brake is smoothed and extended based on the least square approximation. At last, scanning every section plane at a certain interval, on the basis of the back edge of bake, the distant between two points of two edge curves at a section plane is calculated in normal direction. And the least thickness and its position is analysed from all the distants finally.The trials come to that the image grabbing module works efficiently, the images are processed exactly, and the accurate thickness and its position are calculated. Otherwise, the image processing arthimetic is proved to be good stability and repetition and easily adapted to all kinds of situation.