Application Research Of Image Recognition Technology In Ceramic Substrate Grab System

In modern industrial production,with the dramatic increase in production efficiency,traditional manual production methods are increasingly unable to meet the needs of moderm manufacturing.Modern production methods require a new type of automatin technology,and machine vision technology emerges as the times require.Machine vision uses image recognition technology instead of human eyes to make measurements and judgments higher efficiency,better accuracy,more objectivity,and can be repeated indefinitely,which greatly improving the automation of production.In this thesis,the application of image recognition technology in machine vision capture system is studied.The research on ceramic substrate recognition and pose acquisition algorithm is realized,which has certain practical application value.In the research of ceramic substrate target recognition method,this thesis first studies the target recognition method based on Hu invariant moment and KNN algorithm.and based on the shortcomings of the method,the black pixel area ratio,the mean of centroid to edge distance and the variance of centroid to edge distance are proposed as the characteristic parameters of the characteristic parameters.The experimental results show that the latter has small calculation amount,fast running time and can be more accuracy when identify the front and back of the ceram,ic substrate.Because the above two methods can not accurately identify the occlusion of the target object,this thesis further studies the edge-based template matching recognition algorithm,and based on the original similarity metric proposing a improve proposal,which adds the reciprocal of gray value difference as the correction.The experimental results show that the improved algorithm not only can accurately identify the front and the back,but also successfully solve the problem that the target objects are occluded to each other,and is suitable for a variety of complex occlusion situations,which meets the requirements of practical application scenarios.In the research of ceramic substrate target pose acquisition method,this thesis first compares two methods of obtaining the minimum circumscribed rectangle:rectangular envelope method and rotating card method,and calculates the corresponding target centroid coodinate and rotation angle based on the minimum circumscribed rectangle obtained by the two methods.By comparing and analyzing these data,it is finally verified that the method of obtaining the minimum circumscribed rectangle based on the rectangular envelope method and obtaining the target pose data is more suitable for practical application scenarios.Finally,the selected target recognition algorithm and target pose acquisition method are applied in the actual system,and compared to machine vision software Sherlock in the actual application performance.The comparison results show that the target recognition algorithm and target pose acquisition method studied in this thesis is more stable,reliable and robust,and has certain practical application value.