Research On Visual On-line Inspection System For Magnetic Ring Surface Defect

Magnetic ring,a ferrite magnetic ring used in microwave ovens,for the current quality inspection of magnetic rings,it relies on human eye detection,its detection efficiency is low,the quality of detection quality is not high,and a magnetic ring outer surface defect is designed.The detection system,based on machine vision and image processing technology,improves the working efficiency of magnetic ring detection,improves the quality of the magnetic ring,reduces the detection time and finally automates the appearance detection of the magnetic ring in the actual magnetic ring detection.This thesis studies and improves the key technologies of machine vision technology,image processing technology,machine structure control and application software involved in the magnetic ring appearance detection system.Because of the influence of the production process,the magnetic ring will have various defects on the outer surface during the production process.It is necessary to rely on the system to identify and reject the defective magnetic ring.The magnetic ring body is a cylindrical ring,and the outer ring of the magnetic ring needs to be produced during production.The inner edge is cut into rounded corners,which will cause the surface of the magnetic ring to have a grinding texture.At the same time,there will be various defects such as falling blocks,depressions,impurities,mucous membranes,etc.The system needs to detect a wide variety of defects and needs different defects.The magnetic ring is classified.Considering the production cost,it is necessary to judge the pass and fail according to the size and position of the defect,which increases the difficulty of detection.Firstly,the composition and main technical characteristics of machine vision system are summarized,and the application of machine vision in main industries is analyzed.Then,the background and significance of magnetic ring detection are analyzed.According to the requirements of magnetic ring surface defect detection,the structure of the system is summarized.The structure design requirements and design difficulties of the system are studied.Thirdly,we focus on the image processing algorithms of magnetic ring detection,image geometric transformation,SIFT,GLOH and HoG feature transformation preprocessing algorithm,and study the semantic segmentation method based on depth learning,and verify the feasibility and effectiveness of the algorithm.Finally,the system software is developed,and the hardware is designed according to the system,and a prototype is developed.The preliminary test meets the user's detection needs.A large number of field experiments verify the feasibility and reliability of the proposed system scheme and image processing algorithm.