Research On Location Method Of Can Marking Area Based On Machine Vision

In recent years,a way of advertising with physical gifts has sprung up.After purchasing canned drinks,Marketers use laser marking machine to print the company’s advertising information in the blank of the can surface,and transmit it to consumers.This way is novel and unique,and the publicity cost is low,so it is favored by many businessmen.When printing enterprise advertising information on the can body,the main task is to find out the blank area on the surface of the can that can be used for marking.At present,when printing advertising information on the surface of cans,the blank marking area is found by artificial eyes,and it is placed under the marking machine for marking.The accuracy of this way is low,which can not ensure that the same batch of products are the same;Working long hours will cause the operator eyes fatigue,can not complete large-scale production.The automatic location method of blank marking area on the surface of cans based on machine vision is studied in this paper.The following is the main content of its research:According to the technological process of cans marking,the automatic marking equipment for cans is designed,and according to the size and positioning accuracy of cans,industrial cameras and lenses are selected.This paper focuses on the design of the marking area on the equipment,and establishes the image acquisition system of the tank surface.The method of unfolding and stitching of the surface image of cans is studied in this paper.The cylindrical image of the can is expanded into a plane image by cylindrical backprojection.In order to mosaic the can surface image,an image stitching algorithm based on improved template matching is proposed.The algorithm has a fast processing speed and has a good stitching effect.The method of automatically detecting the appropriate marking area in the image is studied.This paper puts forward the idea of binary segmentation of the stitched can surface image by using residual neural network,and automatically searches the blank area that meets the marking requirements according to the binary image,and calculates the rotation angle from the current position to the blank area.The results show that the method can be used to accurately locate the blank areas of different kinds of cans and has strong adaptability.The accuracy of the blank area and the position after rotation are verified by experiments.The results show that the method proposed in this paper can determine the blank area,and after rotating according to the calculated angle,the offset between the center of calibration area and the center of camera field of view is within a certain range,and the positioning accuracy is high,which meets the marking requirements.