Research On Cross-Scale Components Positioning Technology Based On Machine Vision

With the increasing complexity of micro and small products,cross-scale components are more and more common in manufacturing operations.However,at present,the operation process of cross-scale components is basically completed by workers manually,and the success rate and assembly accuracy are completely dependent on individual proficiency,which makes it difficult to meet the requirements of assembly consistency and stability.Therefore,it is of great significance for the development of micro parts manufacturing industry to study the automatic positioning technology of cross-scale components.Firstly,this paper analyzes the cross-scale automatic transfer and positioning system of the functional requirements and design difficulty,on the basis of the overall scheme is determined,the working process of the complete system design and spatial layout,and then transport unit,sample machine unit module for positioning unit and the structure design and the hardware structures,and then according to the layered design way of software architecture design human-computer interaction interface,to realize the state monitoring of each hardware and the control of the upper computer.Secondly,a robot vision positioning and autonomous transport system is designed.Zhang's calibration method is used to eliminate camera distortion,and affine matrix is used to establish coordinate transformation relationship between the camera and the robot.Through a series of image processing methods,such as median filtering,improved Canny edge detection algorithm,rough location of Hough transform,accurate fitting of least square method and so on,the target workpiece can be located quickly and accurately.Then,the obtained position information is transmitted to the controller to guide the robot to identify,locate and transfer the workpiece.Then,a coarse-fine auto-focusing scheme is designed to solve the problems of short depth of field and blurred image of high magnification microscope in the cross-scale components positioning system.Based on the gravity center of the image gray scale is focused on the method of building window,improve the speed and precision of the focus,more comprehensive compare definition function evaluation of the performance,choose different Brenner gradient function and EOG energy gradient function as the coarse fine focusing process definition evaluation function,and then to the shortcoming of traditional climbing method is susceptible to local minima interference is improved,to ensure that the microscopic vision system can accurately capture the focal plane location.At the same time,due to the narrow field of view of the microscope,it is impossible to observe the cross-scale components at the same time,so this paper proposes a method to locate the center of the cross-scale components.Based on the calibration of the relationship among pulse,pixel and length unit,a coarse-fine combination of block search and location strategy is applied to the micro-scale components,and the position and pose of the micro-scale components are obtained by using the minimum encasing rectangle.For large-scale components,the local feature mosaic method based on SURF operator is adopted to conduct panoramic image mosaic.Finally,the central position of large-scale components is obtained by fitting with least square method,which lays a solid foundation for subsequent assembly operations of cross-scale components.