Research On Flange Size Measurement System Based On Machine Vision Technology

Smart measurement plays an important role in smart manufacturing.Accordlingly,smart measurement and inspection based on machine vision technogies attracts more and more attections in smart production fields.The applications of machine vision technologies usually involve the size/dimension measurement of various parts for devices or machinery.Compared with the manual size measurement which is traditionally adopted in factory,advantages of measurement methods using machine vision technologies include non-contact during measurement,rapid measurement with high measurement precision and low cost.Flange is widely used in industrial pipeline system for the purposes of sealing and connection.It has the characteristics of convenient disassembly and high connection strength.But in the processes of part production,due to reasons related to materials’ quality and manufacturing processes,large flange size(s)deviation from the designed ones may happen,and can negatively affect the sealing quality which may cause leakage or even serious accidents.Therefore,it is necessary to measure the key dimensions/sizes of flange.In this study,size measurement system based on machine vision technogies is designed and developed to measure the key dimensions/sizes of flange in the air compressor of one type of air conditioner produced by a well known domestic enterprice.Appropriate algorithms are chosen by comparing the results of different approaches for image processing and analyses.The measurement system developed here can realize the on-line measurement of flange sizes and satisfy the requirements of production.The main research contents of this study are as follows:(1)Size measurement system has been designed and developed for flange.The system consisting of two sub-systems,i.e.,the hardware(physical)system and software system.1)The hardware(physical)system has been designed and constructed by considering the actual production enviorments and manufacturing processes.The core parameters for industrial camera,len and lighting source are determined in order to realize the appropriate lighting conditions and obtain the image with high quality and then the corresponding components are selected.2)Software system has been designed and then developed on Matlab platform.The main works include: a)design and development of HMI(human machine interface);b)the embedment of the algorithm concerning the size measurement and image processing for this study;c)the consideration of image distortion caused by camera and lens;4)the mathematical relationship between one pixel and the value of actual size.(2)The algorithm suitable for this study are determined through analyses and comparsion of the results using different algorithms.The comparsions have been made for all the key steps in image processing such as image gray transform,filtering and noise reduction,image segmentation and pixel level edge detection.In detail they are:a)image grey transformation to reduce the calculation;b)using the median filter operator to eliminate the image noise;c)binary image realization using Otsu threshold segmentation;d)using Canny operator for preliminary edge detection with the precision of pixel level;e)using polar coordinate transformation method to remove the edge of the groove of the center hole for the improvement of the measurement accuracy.(3)Three commonly methods for subpixel edge detection,i.e.,moment method,fitting method and interpolation method,have been studied for the comparisions of their advantages and disadvantages.With the consideration of production requirements,Zernike moment algorithm is selected for subpixel edge detection,and the maximum interclass variance method is applied to Zernike moment algorithm to optimize the selection of step threshold.In this way the improvement of this algorithm has been realized.Then the modified Zernike moment algorithm has been used to extract highprecision flange parts edge.The advantages and disadvantages of Hough transform circle detection and least square circle fitting are compared here.Finally,the least square method has been used to fit the sub-pixel edge of the key edges,and the outer circle and inner hole diameters of the flange are then calculated.(4)The developed measurement system has been tested to verify its functions and to measure the diameters of outer circle and inner hole of flange.The measured values from this system are compared with those from manual measurement,and the absolute error and relative error are obtained and analyzed.The results show that the measurement accuracy of the system satisfy the requirements and the system has fairly good reliability.