Research on defect detection and sorting device of curved label based on multi-template matching

Abstract

Label defects will affect word-of-mouth and sales of goods and bring consumers a bad consumption experience, so it is of great significance to detect label defects. In order to quickly and accurately detect the defects of the curved label and sort the defective products with label defects, the mechanical mechanism and control circuit of curved label defect detection and intelligent sorting device and the label defect method based on visual detection are studied. In the aspect of label defect detection, taking the label defect of a curved bottle as the detection target, using the method of multi-template matching, image acquisition is carried out on the curved label from multiple angles, multiple templates are made, the shape of the image is matched and compared with the established multiple templates in turn, and the highest matching score is taken as the final matching score. The statistical results of 600 surface label defect detection experiments show that this method’s accuracy is 96.5%. Using this method to detect the defects of curved bottle labels has a good effect, and the detection accuracy of curved labels with dense characters and patterns is very high.