With the development of society, enterprises demand higher and higher quality of goods, and label quality inspection system based on image processing is paid more and more attention.

However, in the process of label production, due to the influence of the accuracy of production machines and other factors, there are many quality problems in the labels produced, such as less printing.

Therefore, label defect detection becomes more and more important.

This paper mainly aims at the software algorithm design of label visual inspection system, subtracting the inspection image from the standard image, thus extracting the defective part and solving the problem of less label printing.

The whole process of detecting less printing defects does not need manual work, and this process is handled by the software itself. People only need to confirm the detected tags, so as to realize the perfect combination of artificial intelligence and ensure the quality of tags.