Design And Implement Of An Online PE Bottle Detecton And Sorting System Based On Machine Vision

The modernization production line provides the guarantee for the batch production of filling milk.But in the filling process will inevitably produce substandard products.The prominent problem is the problem of filling rate and density.The traditional method is detecting bottles one by one artificially.This method is labor-intensive,and also false positives and false detection in serious condition.So it is very important to research the automatic defect detection and sorting of online filling dairy products.Accordingly,the detection and sorting of online PE bottles is researched in this paper.The main research contents include:(1)An online PE bottle recognition and localization method based on contour perimeter matching and least square fitting ellipse is proposed.Aiming at the problem of identification and localization of PE bottles with fast motion on the conveyor belt,the collected PE bottle images were pretreated by different filtering and segmentation methods.Choose the median filter and the iterative best threshold segmentation method to extract the feature of the target.Then the contour perimeter matching and the least square fitting method are established to identify and locate the image.The method is effective to complete the detection of PE bottles accurately for its filling rate and sealing property.(2)A PE bottle detection method based on the combine of equal division circle FA and vertical projection method is proposed.According to the detection requirement enterprise of PE bottle filling rate and sealed bottle,with equal circular sealing method to judge the existence of defects.At the same time,a method of measuring the height of liquid level line based on the vertical projection is established.It is effective to detect the moving PE bottle filling rate and tightness.(3)An online PE bottle interception method based on Pyramid iterative method is proposed.On the fast-moving conveyor belt,the traditional PID tracking method is not ideal.In this paper,based on Pyramid iteration is established.Firstly,the transformation relationship between the robot coordinate system,the pixel coordinate system and the transfer belt coordinate system is established.A calibration method based on engineering application is constructed.Then mark the multiple PE bottle targets and calculate the location of the target object on the conveyor belt.Finally,intercepted grasping by the method of Pyramid iterative,It can effectively improve the efficiency of PE bottle grasping.(4)Construction of experimental platform and development of sorting system.The vision system,transmission system,sorting system and other modules are integrated to build an experimental platform.At the same time,the design and research of the communication structure and data acquisition of the discrete system are carried out.And using VAL3 language to write the robot control program and information integration system.The results of the experiment show that the research method and the developed system can effectively solve the problem of defect detection and sorting of online PE bottles.