Research On Surface Defect Detection Method Of Paper-Plastic Composite Bag Based On Machine Vision

At present,the detection of surface defects of paper-plastic composite bags is still based on manual visual inspection,the detection efficiency is low and has strong subjectivity,which can not meet the detection needs of production enterprises.Therefore,it is urgent to study a stable and efficient automatic detection method for surface defects of paper-plastic composite bags.In this paper,on the basis of full investigation of surface defect detection methods at home and abroad,a paper-plastic composite bag surface defect detection method based on machine vision is proposed.The main research contents are as follows:(1)In order to solve the problem of poor applicability of traditional image screening methods for surface defect image screening of paper-plastic composite bags,a method based on gradient projection difference was proposed.According to the gradient amplitude image of the paper-plastic composite bag surface,the corresponding gradient projection mean curve is obtained.By using the "peak" height difference of the gradient projection mean curve between the surface defect image and the surface normal image,a reasonable screening threshold is set to realize the rapid and accurate screening of the paper-plastic composite bag surface defect image.(2)Aiming at the problem that a single segmentation method can’t get ideal segmentation results for a variety of paper-plastic composite bag surface defect images,a new combined segmentation algorithm combining edge detection with adaptive region growing method is proposed.In this algorithm,the centroid of the defect region after edge detection segmentation is used as the initial seed point of the adaptive region growing method,and the flatness of the shape feature is used as the fusion basis of the two segmentation results,the accurate segmentation of five typical defect images on the surface of paper-plastic composite bag is realized.(3)Aiming at the classification problem of the surface defect image of paperplastic composite bag,the shape feature and moment invariant feature of the defect are calculated,and the effective feature selection is carried out.Three support vector machine classifier parameter optimization models based on genetic algorithm,particle swarm optimization algorithm and gray wolf algorithm are established.The effectiveness of feature selection is verified by the test set,and the optimization method of model parameters is determined.The classification accuracy of the optimized parameters training classifier is up to 95.7894%,which is about 5% higher than that of the model before parameter optimization.The accurate classification of paper-plastic composite bag surface defect image is realized.(4)The paper-plastic composite bag surface defect detection platform is designed and built,and the paper-plastic composite bag surface defect detection method proposed in this paper is verified by experiments.The results show that the overall classification accuracy of this method is 96.83%,which has a better detection effect.