Research On Vision Detection Of The Tapered Roller Bearing

Bearing is an important mechanical component and applied widely in the machinery industry. It has a high production and very high requirements of accuracy. Now, most of Chinese bearing factories in dimension measurement for bearing have been mainly dependent on mechanical or optical instruments in sampling inspection way. This method depends on that the human samples survey randomly, which can easily cause artificial error and is inefficient and low reliability. Especially for the tapered roller bearing which exists the problem of inaccurate fitting of rolling element. If its detections also depend on the human, it can easily cause visual fatigue and undetected. Although the method of measuring vibration can detect inaccurate fitting, this way can cause the damage on both the outer ring and inner ring of bearing and increase reject rate.Aiming at these problems, this paper introduces a system which can detect the problem of inaccurate fitting rolling element and dimension measurement based on machine vision technique with the method of image processing technology.According to the detection functions of the system and the characteristics of bearing, the design of detection system for detecting tapered roller bearing has been completed using the technology of image acquisition, display, processing and analysis, which includes the structure design of the system, configuration of hardware, comparative analysis the method of image processing and software design. In the design process, many key problems are solved such as the problem of hardware setting and calibration in image collection system, design of system's light source and lighting scheme. The method of digital image processing and recognition of rolling element inaccurate fitting is studied. The technology related to digital image processing is introduced in detail. In the image preprocessing, this paper analyzes results of several kinds of smoothing filter. Median filter stands out for its good edge maintenance and adaptability. This paper takes median filter to process the collected image and uses histogram modification for image enhancement. Through analyzing and comparing the effect of the classical edge detectors and threshold segmentation algorithm, the paper extracts the image of tapered roller bearing using background correction algorithm, extracts edge information of tapered roller bearing by the method of in gradient on binary image, then fitted the roller bearing's enclose edge by least square method.The paper use the method of template matching in the recognition of rolling element of inaccurate fitting. First, distinguish the top and back of tapered roller bearing. Then recognize the rolling element inaccurate fitting on top and back of tapered roller bearing using method of non-uniform sampling matching and geometric matching separately.A complete vision detection system for tapered roller bearing has been developed based on the platform of LabVIEW by using the powerful image processing software IMAQ Vision. Experiment showed that detection time for a tapered roller bearing met the pipeline requirements, and had higher detection precision. It is proved that it has stronger detection ability.