Research And Application Of Online Detection Technology For Packaging And Printing Defects Based On Machine Vision

In the printing process of packaging paper,there are many types of small defects coexisting,uneven printing surface and random distribution of scratched ink dots,etc.,and the existing industrial production speed is getting faster and faster.The requirements are also higher.With the rapid development of machine vision,replacing manual detection with machine vision detection technology has gradually become a research hotspot.Through the comparative analysis of multiple images,the use of image processing technology for preprocessing,positioning and registration,feature optimization and correction,differential algorithm detection,and multiple types of defects coexisting extraction and other key technologies are studied.The specific research content includes the following points:(1)Machine vision online inspection system platform was built.First,analyze the components of the image acquisition system from the aspects of technical difficulty,image acquisition quality,imaging effect,and data transmission volume,and choose more suitable hardware equipment to build the image acquisition system,aiming to solve the image distortion of the printed matter collected by the camera.,The local quality is blurred,the image is not uniform,and so on.(2)Research on image feature rapid positioning and registration technology.Various noise effects cannot be avoided in image processing,and the subsequent work of the image is reduced with filters.For a single color image containing multiple types of features,a fast small template positioning and registration technology based on the similarity measurement of image information is proposed,which realizes the fast feature positioning and registration.(3)Research on a variety of complex defect extraction technologies.Based on the traditional difference algorithm,the difference improved algorithm is proposed.In order to highlight the difference characteristics,it is processed by image enhancement technology.The experiment found that different trademark papers need different algorithm detection to get better results.In order to eliminate the false edges of defects,the Canny and Sobel operators use the characteristic derivative extremes to eliminate the false edges,and use morphological processing to eliminate the non-defects contained in the image,obtain the real defect map,and extract the entire defect information.(4)Comprehensive analysis of experimental results.A comprehensive test is carried out with four kinds of cigarette small box label paper images,and the advantages of the algorithm detection accuracy,speed and robustness of this paper are demonstrated through experiments.The image defects are classified and the requirements of the actual packaging and printing defects on-line detection are met.