Research On Surface Defect Detection System Of Chip Inductors Based On Machine Vision

In response to the actual production testing requirements of a certain inductance component manufacturer for chip inductors,in order to solve the problems of low efficiency and poor accuracy caused by manual sampling of surface defects in chip inductors,it was decided to optimize the testing method and introduce machine vision testing to improve production efficiency,reduce production costs,and improve the intelligent level of testing.Therefore,this topic will use machine vision technology to achieve the detection of surface defects in chip inductors.Firstly,analyze the advantages and disadvantages of various surface defect detection methods,and select a machine vision detection method for the surface defect detection of the chip inductor based on the actual characteristics of the surface defect of the chip inductor.The overall structure of the visual inspection system is designed,and the entire inspection system is divided into two modules: an image acquisition system and an image processing system.Analyze the lighting effects of light sources and lighting methods,select appropriate lighting schemes,complete the equipment selection of the image acquisition system according to the detection requirements,and build an image acquisition platform.Then,analyze and classify the surface defects of the SMD inductor,preprocess the collected image of the SMD inductor surface defects,first perform camera calibration and distortion correction,then perform image denoising and sharpening,use image enhancement technology to multiply the data,annotate the processed data,divide the dataset,and finally obtain the SMD inductor dataset.Finally,in order to achieve the detection of surface defects in chip inductors,different target detection methods based on deep learning are studied.Based on the DETR model,an improved feature extraction network is proposed to enhance the feature extraction ability of the network model.A spatial self attention mechanism is proposed to enhance the positioning ability of the model object,and shorten the training time of the model.A network structure suitable for surface defect detection of chip inductors is constructed.Based on the improved DETR model,the classification and location of surface defects of chip inductors are realized.