Research On Key Technology Of Intelligent Screw Lock Payment System

In the industrial automation in the process of transformation in our country,a large number of previously used to replace the repeated human labor can no longer be required to meet the current production automation control technology,the efficiency of industrial automation based on rapid development to gradually and AI,big data,more sophisticated information technology such as image algorithm of combining the new industrial age 4.0.And in the production and manufacturing of the workpiece screw lock pay operation is almost inevitable,advancing with The Times,screw lock pay has also gone through the following roughly three stages: manual hand-held lock pay,preset coordinate automatic lock pay,machine vision intelligent lock pay.This paper mainly studies the following two problems of conventional machine vision lock-pay equipment: First question,in gathering for machining screw holes image often meet the camera center with the screw hole is not in the same line,superimposed on the uneven light factors such as edge and center area of the screw holes may produce associated to better separation,for the subsequent image processing algorithm produces large error,It greatly interferes with the correct positioning of the center of the circular hole.The second problem is that the current conventional machine vision center location and circle finding algorithms are mostly based on template matching,Hough transform,least square curve fitting and other methods.In actual production applications,these methods have their limitations and cannot be well compromised with complex application scenarios,cost and efficiency considerations.Aiming at the above two problems,this paper puts forward the corresponding optimization and improvement methods.For the first problem,we optimized the lighting module of the conventional visual locking device structurally by referring to the DOME lighting scheme and the characteristics of the existing structure to adopt a hemispherical backlight lighting scheme.In this scheme,the diffuse reflection of the light passing through the hemisphere-type reflector can illuminate the workpiece evenly,which can distinguish the information of the edge and center area of the screw hole more effectively than the commonly used ring lighting scheme.This scheme not only reduces the difficulty of algorithm optimization but also avoids the complicated operation of binocular camera acquisition.To solve the second problem,based on the classical locating circle algorithm,we firstly cut the contour of the unfixed circle by the grid method,introduced the concept of sample data of the unfixed circle center,and then used the fast Gaussian kernel mean shift clustering algorithm with adaptive bandwidth to approximate the real center coordinates.This method has the wide adaptability that template matching does not have and the efficiency that Hough transform does not have.Compared with curve fitting,this method can directly locate the center of the circle by fitting the contour of the circle first.In order to verify the effectiveness of the improved algorithm,we compare the accuracy and efficiency of the optimized algorithm with the conventional visual center positioning algorithm,and the experimental results prove the excellent characteristics of the algorithm.