Design Of Injection Mold Protector Based On Image Registration And Difference

Injection molding is an important molding process for making plastic products.Injection molded products account for about 37% of the total output of plastic products,and are increasing year by year.The degree of automation of injection molding is high,but in injection production,residues often occur when picking up parts,resulting in mold extrusion residues during the mold clamping process that easily cause mold damage.The traditional mold protection method has a poor protection effect on the mold and cannot achieve predictive protection.Human monitoring of the injection mold condition is also labor-intensive,so the use of image processing technology for residue detection becomes the best way to protect the mold.By consulting the literature and comparing several image processing algorithms applied to the mold protector,it is found that the accuracy of the existing algorithms will decrease significantly when the contrast between the residue and the mold is not high.Therefore,this thesis focuses on the mold protection research of the injection mold residues of the transparent plastic product TPU(thermoplastic polyurethane elastomer rubber)support ring of Baoding Xinke Plastic Products Factory.After understanding the structure and workflow of the injection molding machine on the spot and analyzing the design requirements of the mold protector,this thesis uses a computer-based mold protector design solution,with LabView + VISA as the industrial control software environment,and builds an algorithm development environment based on Microsoft Visual Studio 2019 and OpenCV3.4.1 and OpenCV contrib3.4.1 library functions.In the algorithm environment of this thesis,image registration and image difference are used in combination to realize the detection of TPU residues.First,preprocessing algorithms are used to realize image graying,image enhancement,and image denoising.By comparing the effect and time of SURF and ORB feature detection registration,the SURF feature is selected to be matched by FLANN(Fast Nearest Neighbor Matching),and the 20 best matching feature points are selected to calculate the optimal homography transformation matrix by RANSAC method.The image to be tested is subjected to homography transformation and template image registration.Then the difference between the registered test image and the template image is obtained.After the difference image obtained is intercepted by the region of interest(ROI),binarized threshold segmentation,and open operation,the number of pixels with “1” value is counted to calculate the residual rate.The residual rate value is used to judge whether there is residual in the image.The threshold calculation method based on the standard deviation coefficient is used in the binarization threshold segmentation to enhance the stability of the algorithm.The algorithm was tested by taking 10 images of different transparent TPU support ring residues taken on site.The detection results are all accurate and the algorithm takes a short time.This proves that the algorithm can effectively detect the residue of the injection mold of the TPU support ring.Finally,this thesis selects the I/O board,light source and industrial camera,and designs the front and rear panel programs of LabView and calls the algorithm packaged as a DLL file to complete the system design of the mold protector at a lower cost.Experiments were then carried out in actual production.The experimental results show that the mold protector in this thesis can provide effective protection for TPU injection molds,verifying the robustness of the residue detection algorithm in this thesis,and suggesting some improvements to the shortcomings of the mold protector designed in this thesis.Overall,the application of the mold protector in this thesis can effectively protect the mold,reduce human labor,and greatly improve the automation and intelligence of the injection production process.