Development Of Recognition And Judgment System For Macrostructure Defects Of Continuous Casting Slab Based On Machine Vision

In the process of industrialized continuous casting billet production,cracks,segregation and other defects will inevitably appear in the interior of continuous casting billet.If it is used directly,the service life of the product will be reduced.In order to reduce this phenomenon,it is necessary to effectively monitor the quality of continuous casting billet.At present,the rating and classification of continuous casting billet quality in steel mills still mainly rely on manual experience.This rating method has the disadvantages of inconsistent subjective standards and low detection efficiency,The development of continuous casting slab defect detection system based on machine vision is of great significance for continuous casting product defect detection.Aiming at the macrostructure detection and typical defects of continuous casting slab,according to the process requirements,the image acquisition device for macrostructure defect detection of continuous casting slab is designed,and the hardware detection platform based on machine vision is built;This paper analyzes the defects in the standard Yb / t4002-2003 low magnification structure defect rating diagram of continuous casting billet,obtains the characteristics of typical defects,puts forward the decision tree model method to solve the problem of automatic classification of defects,and the fitting formula guides the automatic rating of defects;Lab VIEW is used to develop the low magnification defect detection system of continuous casting slab,and complete the image acquisition,defect detection and defect classification of low magnification structure of continuous casting slab.Firstly,an automatic image acquisition platform is built based on the principle of machine vision to realize the real-time acquisition of continuous casting slab.Secondly,the low times images of continuous casting billet are preprocessed,and the edge of the image is detected by line detection method,ROI area is acquired,and the noise reduction of image background is realized;The binary method is used to segment the defects,and the extracted defects are morphologically processed to ensure the integrity of the defects.Thirdly,the three characteristics of defect fineness,area and center of gravity coordinate are selected as the nodes of the decision tree,and the decision tree model is constructed by using the maximum information entropy gain method to classify the defects of continuous casting slab;The area ratio and length ratio of defects are extracted as the rating parameters.The rating parameters of defects are linearly fitted with the defect level.The fitting formula can continuously complete the judgment of each defect level,and the obtained defect level can be accurate to 0.1.The average value of correlation coefficient of each defect obtained by fitting analysis is 0.9737,which can meet the accuracy requirements of defect rating.Finally,the software development of macrostructure defect identification and judgment system of continuous casting billet is completed by using Lab VIEW,and the functions of image acquisition,image reading,image processing,image recognition and rating required for defect detection are realized.Through in-depth study of metallurgical standards,this thesis formulates standards for continuous casting slab defect rating,and designs defect detection system,which greatly improves the efficiency and accuracy of low magnification defect detection of continuous casting slab.