The Design Of Wheel Sets Detection Technology Based On Machine Vision

Size of wheel sets is an important indicator to measure the technical conditions. The wheel sets need maintenance after running a period of time, and in the maintenance, the geometry and location of wheel sets relates to the assembly quality and traffic safety directly. Especially in recent years, with the quick development of high-speed railway, the speed, traffic density and load capacity of the train are improving, so the requirement of detection becomes higher and higher.The thesis took the automatic detection instrument of Xinliantie Company as study object, summarized the disadvantages existing in the system, improved the figure treatment system, and made it suffice the requirement of accuracy.For the disadvantages of the system, the thesis summarized the features of the noise and edge in the figures, analyzed the principle and character of several filters and chose the right filter to restrain the noise, analyzed the principle and character of several edge detection algorithms and test them with figures picked by the detection system. And the thesis also proposed an algorithm which used wavelet analysis and Canny at the same time, solved the problem of finding an adaptive threshold. Besides, a method which used 8-connected profile threshold to remove the noise was adopted, then the detection threshold was adjust according to the situation of the noise so that better result could be gained. At last, the points we wanted was standardized with a right way.Through the above work, the thesis solved the problems existed in the system. And using the new system can not only treat the general figures but also test the figure with strong light reflection, and it can judge the situation of the background.