Research On Chocolate Box Defect Detection And Classification Based On Machine Vision

With the development of society and the improvement of people's living standards,people have a higher pursuit of the quality of food packaging.The defect detection of chocolate boxes by traditional human eyes has been unable to meet the requirements of industrial production of chocolate boxes due to low efficiency,substandard accuracy,and rising labor costs year by year.And machine vision detection technology is widely used in industrial production because of its advantages of high detection efficiency and strong anti-interference.Using machine vision to detect chocolate boxes can solve the drawbacks of traditional manual inspection.Therefore,this article takes the chocolate box as the research object,and launches the research on the chocolate box defect detection and classification technology based on machine vision,which has certain practical application value.The main research contents are as follows:First,the characteristics of chocolate box defects and their causes are analyzed.The overall scheme of defect detection and classification of chocolate boxes is designed,the selection basis of each hardware part is studied respectively,and the workflow of software image processing algorithm is analyzed.Secondly,the defect detection algorithm of chocolate box is researched.The image is preprocessed first,and then the maximum class method is used to perform preliminary defect segmentation on the image.The resulting binary image is morphologically operated to remove irrelevant noise and connect the broken edges.Then perform edge detection on the image.The purpose of edge detection is to detect the points with obvious light and dark changes in the image.Through the research and comparison of several edge detection operators.It is found that the result of the edge detection of the chocolate box is not ideal.The detected edge is not complete and lacks adaptability.Based on this,an improved Canny edge detection algorithm is proposed to detect chocolate boxes.The edge detected by this method is more complete and has good adaptability.The accuracy of defect detection can reach97.4%,and the detection time is shorter.It has practical applicability.Finally,the classification algorithm of chocolate boxes is studied.Defect image of chocolate box segmented by threshold segmentation and edge detection.Perform morphological and texture feature extraction.Principal component analysis was used to reduce dimension of extracted features.BP neural network and support vector machine algorithm are used to classify the defect features.The experimental results show that the accuracy of support vector machine algorithm for defect classification is higher than that of BP neural network,the accuracy can reach 96.13%,and the classification is less time-consuming.