Wheel Tread Testing Method Based On Image Processing Technology

The operational statement of wheelset is an important factor that influence the safety of train operation.,so Railway departments always have a strict inpection standard for wheelset.Tread defect of wheelset is a major failure of wheelset,and with the development of high speed and heavy haul trains,the tread defect become increasingly frequently.The wheel tread defect on-line testing method still have many problems,to adapt the development trend of intelligent railway method,the tread defect testing technology should be further studied.At first,the wheel tread defect testing productions at home and board,and the research achievement of wheel tread testing technology were introduced,with briefly analysis of their operation principles and characteristics.Adaptations of these,a new tread defect testing method was designed.In this method,multiple sensors were used to control the cameras,and the control method that can let the cameras catch the whole information of tread was designed.In the aspect of method structure,this paper firstly introduced the hardware structure of wheel tread testing method,and detailed analysed the design principles in method hardware structure.And then,introduced the image processing procedure according to the experiments.Finally,summarized the designed method,and analysed the defections and improvements for them.In details,the paper explained the layout principles of cameras and sensors in tread defect testing method,and the theory of chosing proper parameters for cameras and sensors.In the image processing part of this paper,firstly,we introduced the coarse segmentation method,and the image enhancement technology for color image that used in this paper,and analysed the experiment result with color histogram.Secondly,during the experiments,through observe the characteristics of the images that obtained by the cameras,an improved marked watershed algorithm for color image segmentation was introduced,and explained the procedures of this algorithm according to the experiment.Then,to eliminate the distortion in segementation result,in this paper,a special image rectifying method was designed,and a verify experiment is designed to test the method.Finally this paper introduced a simple method to extract the defects area in wheel tread image,and made some improvement on splitting and merging algorithm.