Injection Molding Raw Material Inspection System Based On Machine Vision

With the development of a number of new industries and new technologies,such as the rise of 5g communication technology,the demand for high-quality injection molded products is also increasing day by day.The injection molding industry is a standard discrete manufacturing industry.The complete industrial chain includes the production,transportation,and storage of raw materials,and then to the injection molding factory to produce molded parts.The quality of raw materials is a key factor affecting the quality of injection molded products,so it is very necessary to strengthen the quality inspection of injection molded materials.The inspection of injection molding materials is a systematic work,and the inspection methods include visual inspection,measurement and trial production.The visual inspection is mainly for the surface defects and dimensional defects of the injection molding materials.However,the traditional manual visual inspection has many shortcomings,and it is easy to cause inspection errors due to the subjective reasons of the workers.This article mainly focuses on the three common pollution of injection molding materials: oxide pollution,organic pollution and heavy metal pollution.The oxide pollution of injection molding raw materials is mainly caused by chemical reactions caused by long-term contact with air;organic pollution is because the resin raw materials have a certain degree of adsorption to the organic matter that may exist in each link of production;heavy metal pollution is generally due to the pollution of the resin storage environment Lead to the occurrence of heavy metal migration.Because these pollutions will cause changes in the color or size of injection molding materials,this article uses machine vision technology instead of manual visual inspection methods.This paper designs the hardware system of resin raw material defect detection and the judgment method of resin raw material comprehensive defect,and optimizes the traditional machine vision method.Specifically,the following work has been carried out:(1)This article designs an injection material inspection system.The hardware design includes designing and integrating hardware parameters such as lighting facilities,cameras and related accessories,and image capture cards.At the same time,a dynamic link library is established through Visual Studio.The designed detection system can complete the high-quality initial image acquisition of the injection material.(2)This article has carried out image preprocessing operations on injection molding raw material images,specifically including image smoothing,image segmentation,morphological processing and sub-pixel edge extraction.In terms of image smoothing,this article compares the effects of multiple smoothing filters and selects a Gaussian filter to reduce image noise;in terms of image segmentation,this article uses dynamic local thresholding to segment ROI;in terms of morphological operations,this article uses the idea of first corrosion and then expansion Deal with the adhesion of injection molding materials;in terms of sub-pixel edge extraction,this paper compares the common operators of pixel-level edge extraction,and chooses Sobel operator as the sub-pixel edge extraction operator,and optimizes the traditional Zernike moment method with non-optimal suppression.As a result of pretreatment,sufficient information about the characteristics of injection molding materials can be obtained.(3)This paper designs comprehensive defect judgment standards for injection molding raw materials based on color and size detection.In terms of color detection,the H component of the image HSV color space is extracted to make the hue histogram of the injection molding material.After determining the ratio of the specific color pixel to the entire image pixel,it is compared with the set threshold;the size detection needs to be based on the sub-pixel edge Perform curve fitting,and the parameters obtained in this process can be used to judge the size of the injection material.Finally,according to the inspection results of the two,the comprehensive defect classification of the injection molding material is carried out.The experiment shows that the accuracy of the comprehensive defect classification is over 80%.