Defect Detection Of Packaging Bag Based On Machine Vision

Packaging bag products play an important role in human life and industrial development,and its quality problems directly affect the safe use of internal products.Traditional packaging bag defect detection uses manual visual inspection to identify,which can’t not only complete this highly repetitive work continuously and stably,but also can’t meet the rhythm of intelligent production.Therefore,this thesis mainly discusses the defect detection of packaging bags based on machine vision technology,and uses robots to retain and eliminate packaging bags,so as to improve the accuracy and speed of defect detection of packaging bags and promote the development of intelligent packaging.Specific research contents mainly include:Firstly,the method of bag surface defect recognition based on feature fusion and semi-supervised cooperative training of random forest is studied.Using Region of Interest(ROI)extraction to reduce redundant information of packaging bags,and extracting edge features by improved Canny algorithm,and extracting HSV color features at the same time,and fusing them to obtain fused features to increase feature discrimination;Then using Random Forest(RF)and Support Vector Machine(SVM)semi-supervised collaborative training,the enhanced sample set is constructed from unlabeled samples,which is used to train RF classifier to improve the recognition accuracy of packaging bag surface defects.Secondly,the thermal imaging defect detection method of packaging bag sealing based on knowledge transfer is discussed.Using thermal imager to obtain the thermal image of packaging bag seal,using small label sample training RF and SVM fusion to build an expert labeling system,marking a large number of unlabeled samples;On this basis,the prediction samples obtained from marking and label samples are combined into enhanced samples,which are input into fine-tuned VGG16(Visual Geometry Group 16)and fine-tuned Res Net 34(Resual Neural Network 34)for training and prediction,so that the recognition accuracy of heat seal defects of packaging bags is improved.Finally,a packaging bag inspection system based on vision is designed,and its architecture is analyzed.The hardware configuration of the system is briefly introduced,and the software interface of the automatic detection system for packaging bag defects is designed.The feasibility of the system is verified by experiments.To sum up,this thesis studies the defect detection of packaging bags based on machine vision,which has certain practical value for promoting the automation and intelligence of industrial packaging equipment.