The Non-standard Plastic Parts Sorting System Based On Machine Vision

In the Internet era where people are always seeking convenience,many platforms of carryout service have emerged,plastic becomes the most widely used material among many materials due to its superiority of tightness and oil resistance and low cost.The non-standard part like plastic take-out boxes are various and the sorting in the production process mainly relies on labor,which has the problems of high cost and unstable efficiency.Therefore,the development of non-standard plastic parts automatic sorting system has important practical significance.According to the characteristics of non-standard plastic parts and their diverse sorting requirements,this article constructs a non-standard plastic parts sorting system with air-suction based on machine vision,which consists of the image acquisition system,the image processing system,and the motion control system.The method and technologies of sorting non-standard plastic parts based on machine vision are researched,and the automatic image acquisition,segmentation,position,classification,and basic defect detection of non-standard plastic parts are realized to support the robot to pick and place the take-out boxes accurately.The main work and research results of this article are as follows:(1)The overall scheme of detection and recognition is designed,the configuration of the image acquisition system and the construction of the experimental platform are completed.The relationship between the coordinate systems is determined by camera calibration.(2)Preprocess the collected image based on visual positioning technology.The original image is denoised with Gaussian filter,then binarized by a threshold-based segmentation method.A contour-based adaptive morphological processing method is proposed to eliminate holes and noise and make the splitted parts adhesive to extract the take-out box accurately.(3)In view of the different actual needs of enterprises,two classification schemes based on appearance and product numbers are studied,and suggestions for different application scenarios are proposed.(4)Through image recognition technologies,the current take-out box is compared with the qualified product to determine whether there are basic problems such as edge defects or overflow.Finally,we conduct experiments on the platform as mentioned above with images of 24 kinds of plastic take-out boxes which are distinguished by their appearance.The results show that the method proposed in this paper has fast recognition speed and high accuracy,and has a wide range of practical engineering prospects and important application value.