Research And Application Of Intelligent Sorting System Based On Machine Vision

With the rapid development of the manufacturing industry,product sorting has become an indispensable part of the production process of the enterprise.Nowadays the sorting objects targeted by the sorting machinery on the market are mostly large and medium-sized objects,but the sorting machinery for thin and light parts is few,such as EMS documents,books,cards,sheet parts,etc.Meanwhile,the sorting methods are asked to make a better improvement.Therefore,the present study is focused on studying a general-purpose sorting machinery for light and thin parts based on image recognition,which has high sorting efficiency and low cost.The present study is focused on studying mechanical device,mechanism motion control and target recognition based on machine vision.A sorting machinery with three-dimensional model of the thin and light parts was designed with reference to the printer and other automatic sorting machinery.The movement state of the light and thin parts is analyzed by calculating the frictional force between the thin and light parts and the roller in the sorting machinery,so that the relationship between the parameters of the light and thin parts and the parameters of the sorting machinery are obtained,which provides a rational discussion for the selection of parameters such as the size of the mechanism.For the control study of the sorting machinery,PLC controller is used to control the sorting machinery operation.Meanwhile,PLC selection is determined by the input and output addresses,and the flow chart is drawn according to the machinery control requirements.The ladder program is written to achieve different goals,in which the sorting machinery can distribute the thin and light parts to the designated sorting port to complete the sorting of the thin and light parts.For the identification of the sorting target,the card identification distribution is obtained.Meanwhile,color recognition,image preprocessing,image correction,image feature extraction,character segmentation on the acquired card image are achieved,completing the extraction of card color,character and pattern features,and then normalize the feature image to obtain feature images with the same size.Finally,the projection statistics method and BP neural network method are used to identify the extracted features in order to complete the recognition of card token information.Based on the above research,the present study attempts to design a bridge deal machine.According to the card parameters,MATLAB is used to solve the parameters such as the size of the sorting mechanism.The bridge deal machine is modeled according to the parameters,and then the model is imported into the multi-body system dynamic simulation software Recurdyn for kinematics simulation.By the analysis of the simulation results,the rationality of the bridge deal machine is verified.Finally,according to the structure of the bridge dealmachine and the PLC ladder diagram,the mechanism action control of the bridge deal machine is designed out.