Research On Theory And Application Of Machine Vision Assembly Robot Based On Template Matching

In the modern production and processing process,the assembly robot is more and more widely used in industrial production lines.It has the characteristics of high precision,high speed,good flexibility,etc.It is driven by AC servo motor and has high repeat position accuracy.It is applied in machinery and light industry.In the automatic assembly,commissioning and handling of related products such as electronics,it is suitable for the flexible automation production needs of modern factories.Machine vision has become an important branch of artificial intelligence development and innovation.It aims to give machines a balance of human vision.Its application and research in the actual work of assembly robots involves the development direction of many formats,and some far-reaching research significance and broadness.Application prospects.In this paper,the application of template matching in the assembly robot vision system is proposed.A template matching algorithm based on improved locust swarm behavior is proposed.So far,almost all template matching methods assume that there is only a strict rigid geometric deformation between the template and the target image,and there is only a translation transformation,so in the case of many differences,such as background confusion and occlusion The outliers are prone to mismatches.Most of the existing algorithms are randomly generated when the initial matching position is selected,which will cause large fluctuations in the matching process,which greatly affects the stability and robustness of the algorithm.Under the prior knowledge of the locust algorithm,a template matching algorithm based on improved locust swarm behavior is proposed.Through the optimization process of template matching,the efficiency of template matching is improved and the interference of noise on the matching process is reduced.Introduce the best similarity detection principle,divide the image into several sub-blocks,and the calculation amount is not too large while the guaranteed point set is large enough,and the size of the sub-block is adjusted accordingly according to the template size.Then,using the pixel positions defined in the specific search area in a given digital image to represent the individuals in the locust swarm,by studying the two characteristics of the locust population in the foraging stage,the locust operator is designed to guidely prevent the locust individuals from prematurely Fall into local optimum.Combined with the matching principle,the matching process is divided into two phases: rough matching and fine matching.Introducing different similarity metrics at different stages,introducing a roulette strategy optimization algorithm to achieve accurate identification and efficient calculation of the position of the target object to be detected.According to the template matching algorithm,the image position information of the target object is identified,the camera calibration and hand-eye calibration are performed,and the actual pose of the target object is calculated by combining the kinematics equation of the robot.The crawling path is calculated based on the shortest path and the most convenient principle of grasping.Get the data file of the robot movement,and finally carry out the crawling test experiment.This paper studies the template matching problem in the assembly robot vision system,optimizes the algorithm structure and the corresponding mathematical model,and forecasts the future research direction of the vision system.