The Application Of Machine Vision In Detection Of The Tapered Roller Bearing

This article is designed to detect system design artificial visual inspection in order tosolve problems in ways to meet the needs of production, for the time allowed todecompose for a general overhaul inner components, we propose a new detection methoduses machine vision to realize tapered roller train the quality of the bearing surface of theloop detection. The entire inspection system is divided into three parts:(1) to obtain high-quality image system architecture design. Machine vision inspection isthrough the collected images were processed and analyzed to find the object to be detecteddefects, so the image quality directly affects the stability and reliability of the machinevision inspection, and even lead to the detection failure. For image acquisition, this paperuse LED light source, using a CCD camera to capture the bearing surface of the loop. Afixed light source and a CCD camera, the use of the motor drive bearings, each part of thebearing for image acquisition and orderly, and save.(2) image preprocessing. The image acquisition process will be affected by variousfactors, resulting in digital images often there are all kinds of noise, noise will cause theimage pixel values obtained do not reflect the true brightness of the scene. In this paper,according to the type of noise distribution characteristics of the relationship, and the workenvironment as well as the determination of the noise image signal, the selection MATLABImage Processing Toolbox provided by linear filtering, median filtering, adaptive filteringand other methods to remove noise, so as to obtain a more accurate image.(3) defect detection program design. Through the collected images were analyzed toidentify, complete the inner raceway is defective detection.Bearing surface is smooth out of the loop a single color, grayscale images thusqualified bearing in its gray values highly concentrated, the damaged bearing damage itssurface smoothness and make major changes in the distribution of gray values occur.Image preprocessed image analysis using matlab image processing toolbox functions, getmean and standard deviation and gray features and good image histogram value of its mean gray values were set a reasonable experimental comparisonand standard deviation, as a standard to achieve defect detection.(4) program design defect classification. The detected defective inner raceway image,image processing to complete its further detailed classification.On the basis of the defect detection procedures for defective image segmentation ofthe original image into a more abstract and more compact form, making a higher level ofimage analysis and understanding as possible. Segmented image edge detection, featureanalysis, based on the geometric characteristics determine the type of defect.