Research On Target Location Technology Of Dispensing Machine Based On Machine Vision

In recent years,with the continuous development of intelligent manufacturing technology,the entity industry represented by manufacturing seeks transformation and upgrading to realize the intelligent development of system equipment.Introducing machine vision technology into manufacturing industry to measure,identify and locate targets can improve production efficiency and reduce production cost.This paper studies the application of machine vision technology in dispensing machine,focusing on the visual positioning technology in machine vision.By realizing a stable and efficient visual positioning algorithm,the production quality and efficiency of dispensing machine can be improved.The main work of this paper is as follows:(1)This paper introduces the hardware structure of the visual positioning system of dispensing machine,and completes the hardware selection by analyzing the parameters and performance of the equipment.(2)Aiming at the problems that the collected image may be disturbed and the amount of calculated data is large,the original image is preprocessed.The noise and impurities that may exist in the image are reduced by image filtering,and the amount of image data to be processed is reduced by image graying,edge detection,image pyramid and other methods.(3)An improved normalized cross-correlation template matching method is proposed.The normalized cross-correlation formula is used as the calculation method of image similarity measure,and the computational complexity is greatly reduced by using Fourier transform and integral graph algorithm.Image pyramid is used to construct low-resolution images.In addition,several intelligent optimization methods such as simulated annealing algorithm,genetic algorithm and particle swarm optimization algorithm are analyzed,and a hybrid swarm optimization method called PSO-GA-SA is proposed to speed up the search process of normalized cross-correlation template matching parameter solution.(4)An improved template matching method based on Generalized Hough transform is proposed.Due to the long detection time and high memory consumption of the traditional generalized Hough transform,the improved generalized Hough transform uses the rotation invariant feature angle formed by any two edge points in the image as the R table index.The 4D voting space composed of traditional coordinates,scales and zooms is reduced to2 D voting space.RANSAC is used to improve the voting process.Combined with the elimination mechanism of in-house points,the number of votes is reduced and the acquisition of matching parameters is accelerated.(5)This paper uses C + + programming language,combined with Open CV image processing library and QT interface development software,designs the dispensing machine visual positioning software,and verifies the feasibility of the algorithm through the positioning experiment of the target workpiece.