Automatic Classification And Encoding Of Industrial Buttons Based On Vision

In today's society,buttons have been widely used in daily life,so the types of buttons are increasingly rich.As it is very low efficient to sort a variety of buttons in the production,the demand of classification encoding button based on intelligent machine vision becomes more and more important.The research of classification of industrial products is mostly using simple texture features and color features.These methods are only suitable for single color or texture industrial product classification.They are difficult to apply to product classification[43]for buttons with a rich and colorful color texture,and it is also difficult to apply to the corresponding product encoding.In view of this situation,this paper combines the template matching and color texture features to classify and encode a large number of buttons with colors and textures.Light conditions and objective factors of collecting button picture caused by objective effect of uneven illumination are difficult to determine,and thus they affect the quality of the buttons pictures.In order to solve these problems,this paper uses the normalization to remove effect of light,and use the adaptive median filter and histogram equalization to enhance image texture.In order to adapt to the changes of the direction,angle and position of the button in the actual production line,the texture feature of the button is extracted by using the gray level co-occurrence matrix which has no distortion.In order to reduce the operation,the gray image template is used,and computational cost of color processing is reduced.Because of the limited training samples,this paper proposes a decision tree to construct the classifier,and then according to the EPC coding rules,the decision tree node is used to encode the button.Due to the type of button over 100,the use of simple color or texture analysis alone is not enough to separate the button,and the simple color and texture analysis method combined with the feature will cause redundancy,which leads to a classification recognition rate is not high.In order to reduce the amount of computation and improve the recognition rate,a fusion method of multi-template matching and color texture feature is proposed for classification and encoding.Experimental results show that the classification method based on decision tree and multi template matching and multi feature fusion can effectively realize the automatic recognition and encoding of the buttons.