Research On Detection System Of Bearing Characters Defects

With further improvement of the speed and scale of bearing production, bearingmanufacturers have given more and more attention at quality. Bearing characters areimportant sign of all bearings, the characters are usually signed on the reference face ofa ring, the non-reference face of a ring, seal, shield etc.. From bearing characters, wecan get the information of bearing model and manufacturer’s name, in some case alsoinclude additional information that is established by manufacturers and users. At present,domestic bearing manufacturers still use artificial visual which is heavy workload andlow efficiency to detect bearing characters, this is extremely contradictory to bearingproduction scale.Based on deep exploration of the theory of machine vision and digital imageprocessing, a detection system of bearing characters defects is designed, the systemsoftware develops in LabVIEW environment which is provided by the NationalInstrument and hardware platform components include the use of red ring low angleLED as light source, the use of Tsugami Corporation35mm lens, SONY XC-ST70CCDcamera and Daheng DH-CG300image acquisition card as input device, the use ofcomputer as processing and output device. The system collects the bearings imagesunder the light source by a CCD camera, the image acquisition card transforms theanalog signal into digital signal and transmits to the computer, and then LabVIEWsoftware platform recognizes the characters, finally the quality of bearing is judged andthe results is displayed.The processing uses binarization, bearing positioning, characters partition,characters normalization, characters thinning to deals with bearing characters image toget the feature of a single character, and then pattern matching. In addition, in the designprocess, aiming at the image characters features, we propose some new methods, suchas rapid Otsu algorithm and improved pattern matching method, the former solves theissues of the bearing characters collected by system which are fuzzy and mutual interference and so on, the latter realizes precise matching of bearing characters.Experiments show that the system has the advantage of interface friendly, stability,short detection time, high recognition rate, can effectively overcome the lack of thebearing characters defected by person. By further, adjusting the height of the camerasystem and the parameters which have been set, the flexibility of system can also detectother models of bearings with characters defects.