Method Research Of Teabag Defects Detection Based On Machine Vision

When a filter paper bag is packaged with tea, people can use it as an integrated entirety-teabag. With the rapid development of society, the pace of life getting faster and faster, the demand of teabag rather than tea is becoming increasingly larger. In the teabag packaging process, there will come some defects on the package. Currently, manual observation is used to detect the teabag defects, while manual observation has some drawbacks such as low detection efficiency, high labor-intensive, instable accuracy of detection, the detection results varies with the enthusiasm and the degree of mental focus of workers.Nowadays, the machine vision technology has been widely applied to products surface defects detection. In this paper, machine vision combined with decision tree classification method is used to classifying teabag defects. The main research contents and results are as follows:(1) Summarize the current research status of machine vision technology, and highlight the current situation of machine vision technology applied in product surface defects detection at home and abroad.(2) Build image acquisition platform of teabags with defects based on machine vision, select the image capture card and camera, and set the arrangement of the light source based on detecting requirements.(3) Introduce the pretreatment of the obtained image. Pretreatment includes filtering and background removal, while the background removal method containing four steps: image segmentation, boundary extraction, seed filling algorithm for obtaining the mask, and removing the image background through the mask.(4) According to the characteristics of teabags defects, process the teabag image with quadrilateral fitting, regional division, and threshold segmentation, and extract the teabags angle character, histogram features and characteristics of the area and so on.(5) Use the ID3 algorithm for constructing a decision tree, select teabags angle character, average gray character of different regions, damaged area features as the decision decisive characters. Use the decisions tree to classify normal teabags, broken teabags, slag inclusion teabags, asymmetric teabags and oblique teabags, the classifying results show that the classifying accuracy rates were 94%, 90%, 86%, 94%, 100% of the above five kinds of teabags, all classification accuracy rates are over 86%, the misjudgment accuracy rates of all the defects were 83.9%, 97.8%, 97.7 %, 92.2%, 94.3%. The reliability of this method was further verified.(6) Use visual C ++ to write teabag processing software interface for teabags defect distribution algorithm demo and classification test. Compared with the manual test results, the software classification accuracy rate was over 86%.