Research And Application Development Of Label Classification And Recognition Algorithms

Product labeling is an important way for customers to understand product information.Incorrect label pasting will not only affect the appearance of the product,but may even cause damage to the reputation of the manufacturer.It will also cause unnecessary trouble for consumers to purchase products.Therefore,it is necessary to detect wrong or missing stickers on the labeling effect.There are more than one label for home appliances,such as energy efficiency,energy labels,parameter nameplate labels,warnings,trademark icon labels,and description and prompt information labels.At the same time,there are also mixed-line production,and the current label detection mainly relies on manual detection,but manual detection has the disadvantages of low efficiency,fatigue,and strong subjectivity,so this paper proposes to use machine vision methods to detect label defects.In view of the above difficulties in household appliance label detection and the actual needs of enterprises,this paper proposes a label classification detection method that generates label template features by analyzing label design files,and designs a set of label detection robot systems in conjunction with 6DOF robots.The system solves the problems of label classification and identification detection on flexible production lines.The main work of this article is as follows:First,analyze the requirements,put forward the problems and methods that the system needs to focus on,and design an overall plan for the label detection system.Then,a label classification algorithm based on design files is designed.The algorithm can analyze the label design draft to realize the classification of the label and the extraction of template features,without the need to establish the template based on the actual captured template image.Subsequently,an online label detection algorithm was designed,which included correction and filtering of perspective distortion and noise to reduce recognition interferenceFinally,the overall function development of the CAT software is completed.Including design file analysis,design manuscript extraction and label position setting function,provide data foundation for core algorithm part,and provide interactive interface for inspection process control.After feature extraction and detection experiments.The effectiveness of label classification and feature extraction algorithms has been verified.At the same time,the robustness,accuracy and real-time performance of the detection algorithm have also been verified.