Design And Implementation Of Automatic Level Bar Inspection And Calibration System

With the development of computer science and technology,machine vision and image processing technology has become a hot research field,and been widely used in industrial production.The level bar is a tool for detection and measurement,and it needs to be calibrated before leaving the factory to ensure the accuracy.Years ago,most level bars are calibrated by human beings,which is inefficient,inaccurate and may cost a lot of manpower and resources.An automatic level bar inspection and calibration system is developed now,and in this system,machine vision and image processing technology is used to detect the position of the bubble in the level bar,calculate the offset of the bubble and determine whether the product is qualified.The system is able to calibrate the unqualified level bars automatically by a motion control card,a servo motor and other equipments.It can greatly improve the accuracy and efficiency,save resources and improve the automation degree of the whole detecting lines.In this paper,we first introduce the background of this topic,the research status and development trend of some image processing algorithms and machine vision system in industrial applications.Then we introduce the hardware structure and software structure of this application,analyse the selection basis of different hardware devices and introduce the software algorithm.We describe the principles and significance of image sampling,introduce the formation of morphological gradients and compare different types of gradient images,analyse the principle of Hough transform.An angle estimation method based on improved Hough transform for this application is proposed,and it has been verified to be feasible and practical.Finally,we elaborate the bubble positioning algorithm.We propose a dynamic view frame generation algorithm based on the graylevel profile of image gradient,which makes the area of the dynamic view frame much smaller than the original image while containing all the information needed for bubble positioning.Then we extract the ROI region from the fuzzy minimized gradient image of the bubble image,and the position of the left and right end points of the bubble is found based on the contour search method based on the run-length coding,and the sub-pixel abscissa of the end points of the bubble can be calculated by Gaussian fitting method.We introduce the calculation method of the sub-pixel abscissa of the horizontal midpoint of the parallel reference lines.According to the geometric characteristics of the bubble image,the method to calculate the offset of the bubble level and the corresponding deflection angle is proposed,which is then converted to the signal required to rotate the motor to calibrate the bubble to be level.