Automatical Encoding of Button Products Based on Visual Recognition

Buttons is closely related to human life The management of buttons in a button factory is very complex because there is a huge number of button types. In order to simplify the management complexity, this paper has proposed a method of multi-template matching to identify the buttons and automatically encode the buttons according to rules in the International Commodity Code so that button libarey can be searched using button pictures. Compared with the traditional template matching, the proposed method makes use of two sets of templates for rough classification, and then uses the texture combined with other features for fine classification, so that the unnecessary matching process can be reduced and the overall recognition rates can be improved to automatically encode button products.