Method Of Dimension Of Bearings Online Inspection Base On Industrial Vision

Bearing is widely applied in machinery industry, and bearing quality concerns the problems of safety production and the usability and lifetime of machine. But manual testing is adopted in bearing inspection, it is not only contribute to heavier workloads and lower efficiency but also seriously affects the exfactorypassrate of bearing caused by the high rate of inspection. So the quality inspection of bearing is one of the problems that should be solved quickly.The paper relies on the project,"Research on inspection and3D reconstruction of micro parts", supported by the Key Technology Program in Shan’Xi province(NO.2011K09-39). A bearing online detection system based on machine vision is designed. This system can simultaneously online detect the size and surface defects of bearing and greatly enhance the detection speed and accuracy of the bearing.Based on the demand of the quality inspections, Firstly, We designs the overall plan of the testing system, discusses the structures of the visual detection’s hardware and its selected principles, especially selects the lamp-source, digital camera, lens and so on, and designs the lighting system after the analysis of the light source and lighting. Secondly, the advantages and disadvantages of different image algorithms are discussed and compared. And the median filter and Iterative method is selected to use in the paper through experimental analysis. The edge detection and feature extraction are especially discussed too. According to the difference between the inspection and defect recognition, different edge detection methods and feature extraction methods are considered in the paper. The detection system proposed in this paper can reduce the amount of computation, shorten the operation time, and greatly improve the precision and speed of bearing detecting. Different bearing types and sizes as the experiment object are detected on the testing platform. The experimental result shows that surface defects and dimension inspection of bearing can be detected quickly and reliably, the measuring accuracy of inner and outer diameter can reach2μm or even higher; the detective surface width of parts is more than20μm. The defects of minimum size more than20μm×20μm can be detected, such as scratch, crack etc. Detection system with defects pattern recognition in the process of image processing can distinguish the type of defect, and improve the detection accuracy; moreover the design of flexible can meet the requirements of different types of bearing detection. It has a good prospect of application.