Application And Research On Machine Vision In Detecting Brakes Of Rail Wagon

Brake shoe has been widely used in domestic rail wagon as a most common way, and will be the important way for a long period. So it is particularly important for effectively and timely detect the status of the brake shoe. At present, the railway department still detect the brake shoe abrasion by technicians. It is has low efficiency, low precision,low reliability, and inconvenient for information management.With the development of image processing and photoelectric techniques, machine vision has been widely used in industry, medicine and so on. It has many advantages like automation, non-contact and on-line, etc. Applied to the detection of brake shoe would be more practical and wider prospect.This paper mainly studies the machine vision technology in the application of the detection of rail wagon shoe. According to the working scene and condition of brake shoe, design and build the shoe image grabbing system. The image grabbing module is made up of lights, lens, CCD cameras and image grabbing, etc; which is used to grab the brake shoe image. The detect system should be calibrate to calculate the thickness of brake shoes: gathering the checkerboard images, extracting the corners, and calculating the corners' image coordinates and physical coordinates, on the basis of the calibration parameter model, finally work out the calibration parameters.According to the feature of brake images, design the image processing arithmetic. Firstly, the original image is cropped by locating the brake shoe area. Secondly, the image is filtered by wiener filter and morphology to reduce the noise and debar the fake edge. Then, by analyzing the character of several edge detecting operators, the Canny edge operator is applied to extract the edge in this process. Finally, According to the distinct feature of the brake shoe edge, this paper devise a set of characteristic vector including continuity length, the lateral variety rate, the relative position between brake and wheel and the distant between two edges of brake. On the basis of pattern recognition theory the two edges of the shoe is extracted. For working out the thickness of the shoe more accurate, the edge is converted to the curve through curve fitting. By comparing the two curves, the paper get the minimum thickness of the shoes and analyses the error. Gathering lots of images to make repeated experiment indicate:the image grabbing module can grab the image with good effect; the image processing can extract the edge of the shoe, locate the position of minimum thickness and calculate the thickness of the shoe. In addition, the image processing arithmetic can meet the requirements of reproducibility and stability.