Research On Micro Dispensing Location And Path Planning Method Based On Visual Recognition

With the microminiaturization of various electronic products and continuous improvement of manufacturing technology,higher demands of encapsulation of such products have been expected by people.During the process of encapsulation,both the volume precision of glue dot of dispensing system and the position accuracy and velocity are all required to be ensured.The volume of glue dot of the micro-dispensing system developed based on the surface tension of tiny droplet in this subject can be up to f L-p L grade.In order to realize the automatic dispensing process of the micro-dispensing system,the dispensing position is accurately and automatically positioned,and the working efficiency of the dispensing system is improved.In this thesis,on the basis of the existing dispensing system,the visual recognition method and path planning method of the target to be dispensed are systematically studied.Firstly,the general design scheme of visual recognition system is determined by combining the features of micro-dispensing system,and the design of hardware system is completed.Open CV image processing technology is used to perform single grayscale processing,median filtering,threshold binarization segmentation,expansion,and other operations on the PCB dispensing circuit board images collected by the system one by one.Then,through the addition of template from 45° and 135°,improvement is made for the traditional Sobel,which was only searching the horizontal and vertical verge,enhancing the detection accuracy of the margin of the dispensing areas.After that,the edges of the fractures are connected by the edge connection algorithm of “edge mode”,and the Hilditch algorithm is used to refine the found repeated edges.Finally,using the minimum circumscribed rectangle of the contour,the centroid of the area to be dispensed is calculated through the relationship of geometric transformation,and the perimeter and area of the dispensing target area are calculated using the point set of the edge contour.Secondly,the objective is to optimize the movement path of the automatic dispensing system and shorten the total stroke of the dispensing system during dispensing.Based on the traditional simulated annealing algorithm,the temperature attenuation method and the third step of the Metropolis sampling algorithm are improved,and the improved algorithm is used to plan the running path of the dispensing system.Compared with the previous algorithm,the operation time of the improved algorithm is significantly shortened.At the same time,this thesis also developed the operating interface of the dispensing visual recognition system,which can display the visual recognition results of the dispensing object and the planned dispensing path in real time.Finally,a dispensing comprehensive experimental platform with integrated visual recognition system was built.In the case of stable environmental conditions,a variety of PCB dispensing circuit boards are used for experimental testing.No matter what kind of PCB dispensing circuit board,the visual recognition system,path planning system,and dispensing system in this thesis can all work normally.The accuracy of visual recognition can reach100%,and the accuracy of recognition can reach 50?m.Among them,the time for the visual recognition system to complete one recognition of the PCB dispensing circuit board for 52 areas to be glued is 247.5ms,the time for path planning is 0.38 s,and the time for the dispensing system to complete the dispensing is 102 s.Finally,the overall dispensing quality of the obtained PCB dispensing circuit board is better.