Method for detecting defects of cosmetic paper label

Abstract

The invention provides a method for detecting the defects of a cosmetic paper label. The method judges whether a label is adhered on a correct position or not, whether the Chinese character identification is correct or not, whether the data is printed on a correct position or not, and whether the date identification is correct or not. A Chinese character extraction algorithm and an identificationalgorithm for different color information of a cosmetic label are designed. A pretreatment algorithm and a high precision identification algorithm for machine printed characters on a cosmetic label are designed. The algorithms are stable, thus the maintenance of the system is convenient, in production, the related control parameters are adjusted, then a start button is pressed down, and then the detection is automatically carried out without artificial monitoring.

Description

technical field [0001] The invention relates to a method for detecting paper labels of cosmetics, belonging to the field of computer vision, in particular to a method for detecting defects of paper labels of cosmetics. Background technique [0002] Labels are used to identify commodities, and have the function of marking product names, models, performance, key technical parameters, instructions for use, prices, ingredients and other key information that consumers and dealers are concerned about. They can be called the "identity card" of commodities. ". Nowadays, many products on the market such as cosmetics, medicines, electrical appliances, stationery, clothing, food, etc. have labels. Labels have become an effective carrier for companies to track products, quality monitoring and production management, and labels are also an effective tool for individuals to check product quality. Especially for exporting foreign trade commodities, the quality of product labels is related...

AI Technical Summary

Problems solved by technology

In this way, there are many limitations and problems: (1) the workers not only need to detect the position of the label and the date, but also check whether the content of the label and the date are correct or not, which leads to a slow detection speed and is difficult to meet the current production speed and requirements; (2) The inspectors perform concentrated work for a long time, which can easily cause eye fatigue, dizziness and other adverse reactions; (3) The inspection results are easily affected by the subjective emotions of the quality inspectors, resulting in the detection of the same product in different inspections. There will also be differences in the eyes of personnel; (4) The artificial detection rate is high, the randomness is large, there is no quantitative standard, and the cost is high

Therefore, not only the requirements for high performance and high resolution are put forward for the hardware, but also the requirements for high stability, high efficiency and strong robustness are put forward for the image analysis and processing algorithms. Accuracy and robustness are difficult to achieve simultaneously

Images