Study On Key Technologies Of On-line Cage Inspection Equipment Based On Image Processing

China has become the world’s third largest bearing manufacturing country,as an important component of the bearing assembly,the output of cage also increased,but it brings pressure on artificial quality testing which is mainly time-consuming and has laborious problems.Machine vision detection technology with its own advantages has been widely used in medicine,industry,agriculture and other fields.Study on key technologies of cage on-line inspection equipment based on image processing can ease the dilemma of bearing industry under the artificial visual inspection.Taking the tapered roller bearing basket cage as an example,this paper researches several key technologies of cage on-line inspection equipment.Main contents are presented as follows:1.Designed the overall composition of the system and complete the hardware selection.According to the existing productivity constraints and requirements of the system,the overall composition of the system is introduced.In the hardware selection,the appropriate components and parameters are selected from the aspects of lighting,camera and lens.2.Completed camera calibration and size calibration.Introduced Zhang Zhengyou calibration method,and extracted corners of different positions of the checkerboard image to solve the camera internal and external parameters.The scale is calibrated by standard parts to convert the pixel values to physical values.3.The sub-pixel measurement method based on Sobel-Zernike moments is proposed.With the Sobel operator,Zernike moment,mask operation and the effective detection,extracted sub-pixel edge points of cage images,complete measurement of the diameter and windows size of the cage.4.Studied the method of measuring the size of the cage in the industrial environment.With the problems of shooting position deviation,image blur and industrial impurity influence,this paper proposes a method based on image center trimming algorithm,sharpness evaluation function,circle searching algorithm and convex hull algorithm.And compared with the results of visual measurement and manual measurement,it is proved that the visual measurement method is feasible.In this paper,the sub-pixel detection algorithm of Sobel-Zernike moments is used to measure the size of the cage,and with some influencing factors in the actual production environment,the measurement method of the cage is given,and it shows the image-based cage size measurement method can replace the traditional manual measurement.